Essays In The Theory Of Employment

TIGHT BINDING BOOK

W > DO ?

OU1 60309 >!

ESSAYS IN THE THEORY
OF EMPLOYMENT

Essays in

the Theory of

Employment

JOAN ROBINSON

OXFORD

BASIL BLAGKWELL
1947

First printed 1937

(Macmillan & Co. Ltd.)
Second Edition October 1947
Reprinted November 1947

Printed in Great Britain for BASIL BLACKWELL & MOTT, LIMITED
by A. R, MOWBRAY & Co. LIMITED, London and .Oxford

FOREWORI

THERE have been considerable developments since these essays
were written, both in theory and in experience. From the point
of view of theory, they belong to the period of the first impact
upon economic thought of Keynes 3 General Theory of Employment,
Interest and Money. From the point of view of experience, they
belong to a period when the existence of unemployment over-
shadowed all economic problems. I think that they are most
easily to be understood in the light of their historical setting,
and that any attempt to bring them up to date by shifts of
emphasis would be confusing. At the same time I believe that
they are not without relevance to the dominant problems of the
present day. I have therefore 'not made any substantial changes
in the text of the first edition.

I have, however, made a few alterations which might equally
well have been made in the first instance. . I have removed
an error from the argument on Mobility of 'Labour (p. 33), I
have simplified the exposition of the effect of inventions on em-
ployment (p. 96), and I have amplified the discussion of the
influence of exchange depreciation on the balance of trade
(p. 143).

JOAN ROBINSON
CAMBRIDGE

February, 1947

FOREWORD TO FIRST EDITION

THESE essays represent an attempt to apply the principles of
Mr. Keynes' General Theory of Employment, Interest and Money to a
number of particular problems. References to the General Theory
are provided for the convenience of the reader, not by way of
acknowledgment to Mr. Keynes, for the very existence of this
book, for what it is worth, must be regarded as an acknowledg-
ment *to the work of Mr. Keynes.
The following are reprinted (each with some alteration) by

vi FOREWORD

permission of the editors concerned: 'Disguised Unemployment
from the Economic Journal, 'The Long-Period Theory of Employ
ment' (except Section 5) from the ^eitschnft fur Nationalokonomie
and the first part of 'Some Reflections on Marxist Economics
from- the Economic Journal. 'An Economist's Sermon' wai
originally delivered to a study circle of the Student Christiar
Movement at Peterhouse, Cambridge.

JOAN ROBINSON
CAMBRIDGE

October, 1936

CONTENTS
PART I

PAOK

FULL EMPLOYMENT - - - , v - - - i

MOBILITY OF LABOUR - - - - - 29

CERTAIN PROPOSED REMEDIES FOR UNEMPLOYMENT 44

DISGUISED UNEMPLOYMENT - - - 60

PART II

THE LONG-PERIOD THEORY OF EMPLOYMENT - 75

THE CONCEPT OF ZERO SAVING - - - 101
DISINVESTMENT - -' - - - -112

DIAGRAMMATIC ILLUSTRATIONS - - - - 119

PART III

THE FOREIGN EXCHANGES - - - - 134
BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOY-
MENT ------- 156

PART IV

INDETERMINACY - - - - - -171

AN ECONOMIST'S SERMON - - - 175

SOME REFLECTIONS ON MARXIST ECONOMICS - 183

PART I
FULL EMPLOYMENT

BEFORE discussing the definition and the characteristics of
full employment; it is necessary to say something about
the factors which influence changes in money wages.
A problem which has caused much perplexity is presented
by the relationship, as it exists in the minds of employees,
between changes in real wages and changes in money
wages. As to what actually occurs there is no dispute.
A cut in money wages will always be resisted by Trade
Unions with whatever force they may command; while a
rise in prices, such as occurs when there is an increase in
effective demand, does not normally lead to the demand
for a rise in money wages sufficient to prevent real wages
from falling. Even when Trade Unions are strong enough
to prevent money wages from falling, they frequently
submit to a fall in real wages, brought about by a rise in
prices and accompanied by an increase in employment.
Now, any one Trade Union, by allowing its money wage
to fall, can in normal circumstances secure an increase
in employment for its members, at the expense of a fall
in their real wage rate; yet a cut in money wages is usually
accepted only under duress. Thus it appears that a fall
in real wages, accompanied by an increase in employment,
is more universally and more strenuously resisted if it is
offered in the form of a cut in money wages than if it is
brought about by a rise in prices. If Trade Unions are
regarded as thinking in terms of real wages their conduct
appears highly inconsistent.

The -explanation offered by Mr. Keynes 1 is that the

1 General Theory of Employment, Interest and Money, p, 14.

2 ESSAYS IN THE THEORY OF EMPLOYM&flT

Trade Unions are concerned, in effect, solely with relative
real wage rates and, within wide limits, pay no attention
to the general level of real wages. While another school of
thought, of which Professor Pigou 1 is an exponent, holds
that the Trade Unions conceive themselves to be stipulating
for a real wage rate, but that, owing to a habit of thinking
too much in terms of money, they fail to realise (except
after a lapse of time) that a rise in prices brings about a
fall in real wages. Each of these explanations appears to
offer an interpretation of Trade Union psychology which
is both dubious and unnecessarily complicated. It seems
simpler to say that since Trade Unions, in the nature of the
case, can only deal in terms of money wages, they concen-
trate their attention upon them, and demand a rise, and
resist a cut, whenever they feel strong enough to do so.
In any actual situation the effect upon employment of a
change in money wages is obscure and difficult to diagnose,
and the suggestion that a cut in wages will do them good
is naturally regarded by the Trade Unions with the utmost
suspicion. Their business is to secure the best terms they
can for their members and to prevent employers from
taking advantage of their monopsonistic position 2 vis-a-vis
the workers. To carry out these functions they must be
on the alert to demand a rise in wages whenever there is
a favourable situation for doing so.

On this view, a constant upward pressure upon money
wages is exercised by the workers (the more strongly the
better they are organised) and a constant downward pres-
sure by employers, the level of wages moving up or down
as one or other party gains an advantage.

In any actual situation employment is likely to ^

1 Theory of Unemployment, p. 294. Mr. Champernowne ('Unemployment, Bar'
and Monetary,' Review of Economic Studies, June 1936) adheres, on the whole, to t/
Pigovian school, but much of his argument is vitiated by a failure to distinguii
between real wage rates and real earnings. Real wage rates fall as employme
increases, but the 'representative* worker (loc. cit. 9 p. 202) must represent both t^.
employed and the unemployed, and the representative worker's real mcome \ i
nominally rise, not fall, as employment increases.

8 See my Economics of Imperfect Competition, p. 282.

FULL EMPLOYMENT 3

increasing and wages rising in some industries ai the same
time as they are falling in others. The present discussion is
confined to a simplified case in which tjiere is perfect
mobility of labour, so that the extent of unemployment
and the pressure upon money wages is the same at any
moment in all trades and all localities. 1 ^

The effect upon bargaining power of a fall in real wages,
in itself that is to say, the effect of 3 rise in the cost of
living relatively to money wages whicn is not a symptom
of an increase in employment is double-edged. On the
one hand, a fall in the standard of life and a reduction in
reserve resources weakens the bargaining position of
workers; on the other hand, indignatipn both cements the
bonds of Trade Union organisation and makes the demand
for higher wages more insistent. 2 Thus a rise in the cost of
living, in itself, may lead either to a rise or to a fall in the
level of money wages according to the strength and policy
of the Trade Unions.

An increase in effective demand, given the state of
Union organisation and the general conditions of the
labour market, will be favourable, for several reasons, to a
rise in money wages. First, a man who is secure of his
job suffers a loss of real wages when effective demand
increases and the cost of living rises; he consequently
becomes more anxious than before to demand a higher
money wage. A man who is out of work or who holds a
precarious job would prefer not to open the question of
money wages and to allow the increase in employment to

1 Some problems connected with imperfect mobility are discussed below, p. 29.

2 Similar contradictory tendencies are seen at work in the determination of
vomen's wages. It is commonly said that women's wages are relatively low because
iey have no family obligations. In itself the carefree state of women should improve
icir bargaining position, since unemployment for them is less to be dreaded than it

for men. But the fact that their needs are less reduces the incentive to organise,

.akes their demands less insistent, and deprives them of the backing of general

iblic opinion, with the result that their bargaining power is generally weaker than

at of men. The relatively backward state of women's organisations is not wholly

be accounted for in this way, but is also partly due to the fact that a woman does

l t normally look forward to spending her whole life in industry, and so has less

^tive to demand improvements in the conditions of employment, and, we are

to the lack of co-operative spirit characteristic of the female sex.

4 ESSAYS IN THE THEORY OF EMPLOYMENT

take its course. As unemployment falls off, men of the first
type increase, and of the second type decrease, in numbers
and influence. A Trade Union which takes the interests of
both types into account will therefore be gradually moved
in the direction of demanding higher wages as unemploy-
ment falls off. Second, the existence of unemployment
weakens the position of the Trade Unions by reducing
their financial resources and awakening the fear of com-
petition from non-union labour. Thus even a Union
which at the moment represents only employed workers
will be more restrained in its action the greater the amount
of unemployment outside. Third, the strategic and moral
position of Trade Unions is strengthened when profits are
rising and real wages falling. Conversely when effective
demand declines. Thus movements of the level of employ-
ment are the chief influence determining movements in
the level of money wages, j

It is idle to attempt to reduce such questions as Trade
Union policy to a cut-and-dried scheme of formal analysis, 1
but it is plausible to say, in a general way, that in any
given conditions of the labour market there is a certain
more or less definite level of employment at which money
wages will rise, and a lower level of employment at which
money wages fall. Between the two critical levels there
will be a neutral range within which wages are constant.

It is sometimes argued that movements in money wages
tend in the long run to eliminate unemployment, for a fall
in money wages, by reducing the demand for money,
tends to lower the rate of interest, and so to increase
employment. 2 ^ But the most that can be hoped, in a
community where Trade Unions exist, from a policy of

1 Cf. below, p. 171.

2 General Theory, p. 263. Cf. below, p. 86. It is by no means a corollary of this
argument that the level of employment would tend to be higher in a community
in which there were no Trade Unions to resist a fall in wages when unemploymeni
occurs. For without Trade Unions monopsony is more powerful, the ratio of rea
wages to real profits is smaller, thriftiness is greater and the level of employmer
with a given rate of interest consequently lower (see p. 94) . This effect will te
strongly against the influence of low money wages in increasing employment.

EMPLOYMENT 5

maintaining a constant supply of money and allowing
wages to find their own level, is that employment shall be
maintained inside the neutral range within which money
wages are constant, which is by no means the same thing
as the point at which unemployment is tending to dis-
appear.

When employment stands above the upper critical level,
then, if conditions are such that a general rise in money
wages sets up no reaction to reduce effective demand,
there will be a progressive rise in wages with a constant
level of employment, for prices and profits will rise with
money wages and all the circumstances which led to a first
rise in wages will remain in force and lead to a second. 1
But the existence of unemployed workers anxious to find
jobs exercises a drag upon the Trade Unions, and the rise
in money wages will be slight and gradual. An increase in
employment, in this situation, will strengthen the Trade
Union position and tend to speed up the rise in money
wages, but so long as unemployment remains appreciable
the upward movement cannot become overwhelmingly
powerful. Thus there is no paradox in the fact that effec-
tive demand may increase, and real wages fall, in a situation
in which the Trade Unions are both able and willing to
prevent a fall in money wages, or even to procure a rise.
The connection between movements in money wages and
movements in real wages is largely accidental. There is a
certain level of employment, determined by the general
strategical position of the Trade Unions, at which money
wages rise, and at that level of employment there is a cer-
tain level of real wages, determined by the technical
conditions of production and the degree of monopoly. In
demanding a rise of money wages the Trade Unions are

1 Our formal scheme, in terms of the critical levels of employment, involves a high
degree of simplification, for the movement in wages will not only depend upon the
absolute level of employment but also upon recent changes in it. It is reasonable to
say, however, that the upper critical level is such that, so long as employment stands
appreciably above it, a small fall in employment will give only a temporary check
to the rise in wages, and that when employment has continued for some time at its
reduced level the tendency for rising money wages will resume its force.

6 ESSAYS IN THE THEORY OF EMPLOYMENT

not taking a view upon how much unemployment they
want to have, but merely seizing advantage of the fact
that unemployment is low. The paradox of Trade Union
policy only appears when we regard the Trade Unions as
thinking in terms of the real wage corresponding to a given
level of unemployment, instead of regarding movements
in the cost of living as merely one element in their bar-,
gaining position. The economists have created a puzzle
for themselves by inventing the answer of the Trade
Unions to a question which they do not

v Under conditions of perfect competition with no labour
organisation and perfect mobility, there is no difficulty
in defining full employment. In these conditions the two
critical levels coincide at the point of full employment.
So long as any unemployment exists money wages are
falling, and when full employment is reached they rise
abruptly .v The movement of money wages then serves to
indicate the existence of unemployment. But when workers
are prevented, even by tacit agreement, from under-
bidding each other for jobs this criterion fails.

It would, indeed, be permissible, from one point of view,
to say that there is no 'involuntary' unemployment as long
as unemployed individuals are not so anxious to find work
that they are prepared to offer themselves at cut wage
rates. But this use of terms would be extravagant. From
a common-sense point of view it would be absurd to say
that there was no involuntary unemployment in Great
Britain in 1933 because at that date money wage rates
were stable, or that there was no unemployment in coal
mining in 1936 because a demand was then made for
higher money wages. If the term 'involuntary unemploy-
ment 5 is confined to the sense of a degree of unemployment

1 A Trade Union policy conceived in terms of real wages is discussed below, p. 27.

2 The conception of full employment where mobility is not perfect is discussed
below, p. 41.

EMPLOYMENT 7

SQ severe as to break through the feeling of loyalty and the
Trade Union regulations which forbid money wages to be
cut, then some other term must be found to correspond
to the common-sense notion of unemployment. But the
term 'involuntary unemployment' is obviously suitable
for the latter purpose, and it seems most satisfactory simply
to say that ^the amount of involuntary unemployment is
the amount of work which, in existing conditions, the
population is willing but unable to perform." No absolutely
precise measure of this quantity can be found, but for all
practical purposes a common-sense definition will suffice.
In this sense unemployment may exist even when money
wages are rising/'

Mr. Keynes describes the point of full employment in
terms of the level of real wages. In his view, as effective
demand increases and real wages fall, a point is reached
at which the real wage rate is so low that workers refuse
to supply labour at a lower Yate. At this point the expan-
sion of effective demand must come to an end, because
beyond it labour refuses to go. This conception involves
certain difficulties. <sOn the one hand, the individual
breadwinner without private means can never be in a
position to refuse to work because real wages are too low
to be worth the effort? He must earn what he can get
or starve altogether. Even if he could retain his right
to the dole after refusing a job at the ruling wage rate, he
would find that the real dole had fallen as much as the real
wage. tfOn the other hand, a Trade Union will threaten
to withdraw its labour, not when any particular level of
real wages is reached, but whenever the general situation
(in which the level of real wages is only one element)
appears to offer a favourable opportunity for doing so>
^It is true that pin-money girls may not be lured intc
the labour market by the offer of a wage that will buy less
than so many pins^UUs true that if workers are free to decide
their hours of worlcfhey may refuse to work an additional
hour unless the additional earnings seem worth the effort.

8 ESSAYS IN THE THEORY OF

and it is true that the same individual may work wit,h
greater or less intensity in different circumstances. But
there is little support to be gained from these facts for
Mr. Keynes 5 point of view\ For it is commonly found
that hours become longer and the number of workers in a
family greater as real wage rates fall. 1 In short the supply
of labour from a given population is likely to have a nega-
tive, not a positive elasticity, in response to changes in real
wages$ up to the point at which the maximum limit of
physical capacity is reached .3 We may simplify our argu-
ment, without sacrificing any point of principle, gtf we
postulate that the supply of labour has zero elasticity so
long as real wages stand above the level at which physical
efficiency is impaired by a low standard of life. In this
case there is a certain stereotyped length and intensity of
working day and a certain number of individuals who
desire to have jobs^

When all these individuals are in employment there
is no longer anyone whose interest lies in refraining from
asking for higher money wages when a further rise in
prices takes place. The fear of blacklegs disappears, and
the position of the Trade Unions is extremely strong.
There is no need to postulate a critical psychological point
at which the real wage ceases to cover the disutility of
labour.Ot is true that as real wages fall the demand for

1 Cf. p. 120.

1 If the supply of labour is not perfectly inelastic the amount of labour available
to be employed varies with the rate of real wages. If the elasticity is negative the
amount of available labour increases as the amount of employment increases and real
wages fall. Full employment obtains when an increment of employment, if it were
attained, would be associated with a fall in real wages insufficient to induce an equal
increment in the available supply. When the elasticity is positive the available supply
of labour falls as real wages fall, and, in a sense, unemployment decreases faster than
employment increases. But these complications introduce no difference in principle
from the simple case in which the elasticity of supply is zero. Cf. p. 132.

8 The unsatisfactory nature of Mr. Keynes' psychological interpretation of full
employment is clearly seen if, anticipating the argument of the next essay, we remove
the assumption of perfect mobility and consider the case where full employment is
reached for a particular grade of labour, for instance, master-plasterers during a
building boom. When this occurs the money wages of that type of labour rise ahead
of the general level and their real wages increase. The resistance which then arises

f> FULL EMPLOYMENT 9

Jjigher money wages is strengthened, and it is true that,
as effective demand increases, the threat of a Trade Union
to withdraw its labour becomes more powerful; But this
is merely a matter of degree. The demand for higher
money wages is made with growing success as employment
increases and when full employment is reached it becomes
overwhelmingly strongs The point of full employment
is the point at which 'every impediment on the side of
labour to a rise in money wages finally gives way. ,

Moreover, when full employment obtains, the attitude
of employers to a rise in money wages is radically changed.
In this connection it is necessary to distinguish between a
state of affairs in which the level of effective demand
happens to be exactly that which gives full employment,
and a state of affairs in which effective demand is prevented
from expanding only by the fact that full employment
has already been reached. '/In the first case the demand
for labour is exactly equal to the supply. In the second
case there is a scarcity of labour, in the sense that the demand
for labour at the ruling money wage rate exceeds the
available supply. The extent of the scarcity may be
measured, in a rough way, by the extent to which employ-
ment would have increased if the supply of labour were
unlimited^

Now, when there is a scarcity of labour, employers
themselves have an incentive to raise wages. On the
assumption of perfect mobility, full employment will be
reached in all trades at the same time, and when there is a
scarcity of labour each employer will be in a situation in
which he could increase his profits if he were able to secure
more workers, even at a somewhat higher wage. But more
workers are only to be had by tempting them away from
other employers. A strong convention prevails that it is
a dastardly act for one employer to lure away labour from

to a further expansion in the output of plastering clearly cannot be expressed in terms
of disutility of labour. It arises simply from the fact that all the labour of that par-
ticular type is already in employment.

io ESSAYS IN THE THEORY OF EMPLOYMENT

another by the offer of higher wages, and so long as thi$
convention holds in full force the attitude of employers to
a rise in wages is no different when there is full employment
from their attitude when there is unemployment, but in
the face of a considerable scarcity of labour the convention
is likely to break down, and employers, each following the
dictates of self-interest, will begin to bid up money wages.
When labour is scarce not only are Trade Unions very
powerful, but employers themselves throw their weight
into the scale of rising wages. Conditions? of full employ-
ment obtain when no one employer can increase his staff
without reducing the staff of some other employer.

3

It is clear that, when the point of full employment
has been reached, something must occur to prevent
employment from expanding any further, but it is not
immediately obvious how this comes about. At first
sight it appears that there is nothing to prevent, for instance,
the rate of interest from being set so low, or inventions
from occurring so fast, that the appropriate amount of
employment lies beyond the limit set by the available
supply of labour. ^

There is no corresponding problem in connection with
the limit set by capital equipment. For if equipment is
limited, prices rise faster than money wages as output
expands, and real profits increase as real wages fall.
Saving from profit incomes exceeds saving from wage
incomes: thus, as output expands, the ratio of saving to
consumption increases, and it will increase faster the more
inelastic is the supply of output from given equipment.
In the limit, if no increase in the output of consumption
goods is technically possible, prices (and consequently
profits) will rise to whatever point is necessary to prevent
consumption from increasing. 1 But when the limit is set

1 There appears to be no warrant for Mr. Hicks' view (Economic Journal, June 1936,
p. 247) that an indefinitely great rise of prices will develop in a state^where capacity

iFULL EMPLOYMENT 11

by labour, instead of by equipment, this check upon the
expansion of effective demand does not operate.

If employers are in a position to maintain the convention
against raising money wages, the situation is very similar
to that in which output is limited by the capacity of plant.
In this case, if, for instance, the inducement to invest is
increased when full employment already obtains, output
is held in check merely by the impossibility of finding
additional workers. The price of capital goods rises, but
their cost is unchanged. An increased monopsony profit
therefore falls to the investment entrepreneurs. 1 If part of
this profit is spent, prices of consumption goods also rise
above their cost, and the rise of prices, for all goods, will
proceed to the point at which increased saving out of
profits is equal to the increased value of the constant out-
put of capital goods. In short, the increased inducement
to invest automatically brings about a reduction in the
propensity to consume, due to increased maldistribution
of income, which keeps the total demand for labour un-
changed. But when Trade Unions are strong, and when
self-interest breaks through the class loyalty of employers,
this check upon output also fails to act.
t It is therefore necessary to inquire what mechanism
there is, in the general case, which prevents the economic
system from overshooting the point of full employment.
To illustrate this problem let us suppose that, when full
employment already obtains, the rate of interest is reduced
and is maintained continuously at the lower level. Our
discussion will be confined to the simplified case in which
there is perfect mobility of labour and no element of
monopsony in the labour market.

is limited but available labour unemployed. It is true, as Mr. Hicks says, that when
effective demand increases and the supply of money is constant, the rate of interest
will rise if the supply of consumption goods is less than perfectly elastic, so that prices
rise. But it will rise even if supply is perfectly elastic and prices are constant.

1 Similarly, when special rates are paid for overtime, monopsony profit increases.
For the marginal net productivity of labour is at least equal to the overtime rate, and
is consequently greater than the average wage rate (cf. Economics of Imperfect Competi-
tion, p. 301).

12 ESSAYS IN THE THEORY OF EMPLOYMENT

^ The reduction in the rate of interest raises the price of
capital goods, and under the incentive of increased profits
a demand for labour is set up in the capital-good indus-
tries, 1 but every available man is already employed. This
must lead immediately to a rise in money wages in the
industries engaged upon capital goods. 2 Expenditure from
these wages increases the money demand for consumption
goods, but in consumption-good industries also full em-
ployment already obtains. An equal rise in money wages
in the consumption-good industries is necessary to prevent
a transfer of labour into the capital-good industries. Each
individual employer in the consumption-good industries
finds that the demand for his product is increased, and is
reluctant to lose labour. The rise in money wages there-
fore communicates itself to all industries. )

The manner in which the situation develops will depend,
first, upon the way in which the change in wages is com-
municated from industry to industry, and second, upon the
way in which the expectations of entrepreneurs adapt
themselves to the changes in wages. Let us assume for
the moment that the changes in wages are instantaneous,
so that a rise in wages in the capital-good industries is
transmitted to the consumption-good industries without
any actual transfer of labour between the two groups.
In this case no increase in investment, reckoned in wage
units, can take place.

The expectations of entrepreneurs may be divided into
two types: expectations as to the future course of money

1 To avoid unnecessary complexities it is convenient to assume that the change
in the rate of interest has no reaction upon the amount which individuals desire to
save out of a given income. When this assumption is not fulfilled the demand for
labour in consumption industries also will be altered.

2 A clear-cut distinction between capital -good industries and consumption-good
industries is not found in reality. Any industry is engaged in investment in so far as
its current output is not currently consumed, and goods partake of the nature of capital
goods to the extent that their price is affected directly by a change in the rate of
interest. In this respect all goods can be arranged in a continuous series. But for verbal
simplicity it is convenient to talk as though industries could be divided into two self-
contained groups. The fact that the two groups are not completely self-contained
makes no difference of principle to the argument.

^FULL EMPLOYMENT 13

wages, and expectations as to the effect 01 a given change
in money wages upon future prices. Let us first suppose
that entrepreneurs always expect that to-day's wage rates
will remain in force in the future, and consider the effect
of expectations of the second type.

As soon as a rise in money wages occurs the entre-
preneurs engaged upon the production of capital goods
find that their cbsts have risen, and the incentive to invest
thereby tends to be reduced. But the prices of the com-
modities which ,the capital goods are designed to produce
have risen in the same proportion. Thus, if entrepreneurs
adjust their expectations of the future earnings of capital
goods instantaneously to the change in wages the marginal
efficiency of capital is unaffected and remains the same
as it was before the rate of interest fell. Thus there is no
possibility of the marginal efficiency of capital being
reduced to equality with the rate of interest, either by an
increase in the rate of investment (reckoned in wage units)
or by a rise in cost of capital goods relatively to their ex-
pected earnings, and no equilibrium is possible. J

But it is natural to suppose that entrepreneurs react
more quickly to a rise in the cost of capital goods than
to the change in prospective earnings due to a rise in the
price of consumption goods, for the change in cost presents
itself to them more immediately. A speculative builder
becomes aware of a rise in the cost of building more
rapidly than of a rise in house rents, even though the two
proceed at the same rate. ( An all-round rise in money
wages will thus reduce the marginal efficiency of capital
for a certain time after it has taken place, the length of
time depending upon the rate of adjustment of expectations.
If the pace of adjustment is given there is one rate of rise in
wages which will preserve equality between the marginal
efficiency of capital and the reduced rate of interest.
But if, at any moment, the marginal efficiency of capital
is greater than the rate of interest, the competition of
entrepreneurs for labour must at that moment be driving

14 ESSAYS IN THE THEORY OF EMPLOYMENT

money wages upwards. Thus< the rate at which money
wages rise will necessarily be whatever is required to main-
tain equality between the marginal efficiency of capital
and the rate of interest. The more rapidly are the expecta-
tions of entrepreneurs revised the more rapid will be the
rise in money wages. Expectations will be revised more
rapidly as experience teaches the entrepreneurs that a rise
in costs in their own industries is accompanied by a rise
in prices in others. Thus the rate of rise in money wages
will accelerate as time goes by. In the lijnit, as we have
seen, if expectations are revised instantaneously money
wages must rise indefinitely fast.

^We have so far considered expectations in the minds oi
investment entrepreneurs as to the effect of a rise in
wages, which has already occurred, upon the future prices
of consumption goods. We must now consider expectations
about the future course of wages themselves. It is possible
that the rise in money wages may set up an expectation
that they will fall again in the future. Such an expectation
would partially or completely offset the effect upon the
inducement to invest of the initial reduction in the rate oi
interest. Thus an expectation that wages will fall in the
future retards their rise in the present. But since there is
nothing in the situation actually to bring about a fall in
wages the expectation of a fall will weaken as time goes by,
and as the inducement to invest recovers the competition
for labour will once more set in. Experience will soon
teach the entrepreneurs that when labour is scarce a rise
in wages is likely to lead to a further rise, and when their
education has reached this point each rise in wages will
enhance the inducement to invest and so precipitate a
further rise. Thus there is an additional reason to expect
that the upward movement will be slow at first but con-
tinuously accelerating.

On the assumption that changes in money wages are
communicated instantaneously from industry to industry,
equilibrium is preserved, with rising wages and prices,

FULL EMPLOYMENT 15

at the point of full employment, and the rate of investment
{reckoned in wage units) does not increase when the rate
of interest falls. ^ If wages are not adjusted instantaneously
there may be some transfer of labour between industries.
Let us suppose that wage bargains can only be altered by
discrete steps and at discrete intervals of time. When the
rate of interest falls the inducement to invest is increased
and employers in the capital-good industries offer a rise
in wages in order to attract more labour. The employers
in consumptipn^good industries may be supposed to lag
behind, and for one bargaining period they must submit
to losing part of their labour force to the investment
industries. Investment reckoned in wage units will there-
fore increase. The additional investment incomes will be
partly spent upon a now diminished supply of consumption
goods, and profits will rise to the point at which there is
an addition to saving equal to the addition to investment.
But the original rate of output of consumption goods was
in equilibrium with the original rate of investment. Now
investment has increased and the profitability of the original
output of consumption goods has been raised. Entrepre-
neurs producing consumption goods are therefore under
an incentive to attract back into their industries the labour
that they have recently lost. If they succeeded in attracting
it back again investment would fall off, and the surplus
profits due to a scarcity of consumption goods would
disappear. But now investment would have been forced
back below the level at which the marginal efficiency of
capital is equal to the reduced rate of interest, and the
investment entrepreneurs would once more endeavour to
tempt labour away from the consumption entrepreneurs.
Thus we may either imagine that there is a perpetual tug-
of-war between the two groups, each suffering a contrac-
tion when the other succeeds in expanding, 1 or we may

1 This analysis bears a superficial resemblance to that of Professor Hayek (Prices
and Production, Lecture III). But his attempt to discuss the course of the trade cycle
upon the assumption of full employment leads to many difficulties, and it appears
impossible to find any genuine point of contact between his argument and the above.

16 ESSAYS IN THE THEORY OF EMPLOYMENT

suppose that the investment entrepreneurs, after snatching
an initial advantage, retain it by keeping the wages that
they offer always a little ahead of those offered in the
consumption-good industries. In either case a continuous
rise in money wages must occur, each successive bargaining
period leading to an increase in the bids of the rival
industries for labour.

We have so far ignored the effect of contracts fixed in
terms of money. /As money wages and prices rise the
purchasing power of incomes fixed in t^jrms of money is
continuously reduced, and real profits are swollen to a
corresponding extent^ If the expenditure of entrepreneurs
is increased by less than the expenditure of rentiers is
reduced, some labour will be released from the consump-
tion trades and can be absorbed in the investment trades. 1
Moreover the labour supply may be augmented by rentiers
who are forced to start earning to supplement their vanish-
ing real income. It is conceivable that the labour made
available in these ways might be sufficient to meet the
increase in demand in the capital-good industries set up by
the initial fall in the rate of interest. In that case the level
of money wages would come to rest after a certain point.
In other words^the rise in money wages may reduce the
propensity to consume of the community, and may in-
crease the amount of labour corresponding to full employ-
ment, and this may occur to a sufficient extent to allow
such an increase in investment (reckoned in wage units) as
will reduce the marginal efficiency of capital to equality
with the rate of interest, even if money wages become
constant"^ Against this possibility must be set the fact
that after prices have been rising for a certain time an
expectation is likely to be set up that they will rise further
in the future. This, as we have already seen, will further
enhance the inducement to invest, since it will lead to a
rise in the present price of all durable goods over and

1 It is this phenomenon which provides the most precise meaning which can be
attached to the much abused phrase 'forced saving'; cf. General Theory, p. 80.

EMPLOYMENT 17

the ris^initially caused b^: the fall in the rate of
interest. But it will edso have the effect of reducing the
incentive to save, since the amount of consumption, sacri-
ficed in the present by an act of saving, is greater than the
amount of purchasing power made available by it in the
future. The purchasing power of money over perishable
goods is expected to fall in the future, and purchasing
power over durable goods has fallen in the present. 1 Thus
it is far more likely that, on balance, the propensity to
consume will be increased than that it will be diminished.
In so far as it is increased the competition for labour of the
consumption trades becomes hotter and the pace of the
rise in money wages all the more violent.

Our analysis has been conducted upon the unreal
assumption of perfect mobility of labour between trades
and localities, but it is clear that the main conclusion,
that conditions of scarcity of labour are calculated to lead
to a rapid and accelerating rise in money wages, is likely
to hold good on more realisitic assumptions. The general
upshot of our argument is that the point of full employ-
ment, so far from being an equilibrium resting place,
appears to be a precipice over which, once it has reached
the edge, the value of money must plunge into a bottomless
abyss. But our discussion has been based upon the assump-
tion that the initial cause of an increase in effective demand
retains its force when money wages rise. This assumption
must now be revised.

A rise in prices and incomes leads to an increase in
requirements for money balances in the active circulation.
This tends to reduce the amount available for inactive
balances and so causes the rate of interest to rise. 2 The case
which we have been considering, for the sake of illustration,
in which a reduction in the rate of interest is maintained

1 In short, the 'real* rate of interest has fallen. 2 See General Theory, p. 171.

B

1 8 ESSAYS IN THE THEORY OF EMPLOYMENT

in the face of a considerable rise in money^wages is not a,
case which can normally occur. Before the'value of money
has bounced more than once or twice against the wall of
the cliff as it falls, a rise in the rate of interest draws taut
the rope that holds it and brings it to rest. VTh_ ri se * n
interest checks investment and continues to the point at
which sufficient unemployment occurs to prevent a further
rise in wages. Most often the movement of wages is
reversed by the reaction, and the laborious task of hauling
the value of money up again begins tq be carried out.
The rope which holds the value of money is a limitation
on its supply. This may be due to natural causes (as in a
primitive metallic system), to legal restrictions, or to the
deliberate policy of the monetary authorities.

This safeguard acts even in a closed system. In an open
system it is reinforced by the effect of rising money wages
upon the balance of trade. \A rise in wages in any one
country will weaken its international competitive position,
lead to a decline in exports and increase in imports and so
tend to cause unemployment. The release of labour from
the industries producing tradeable goods will have a direct
effect in checking the rise in money wages, and the decline
in the balance of trade will have an indirect effect, by raising
the home rate of interest. 1 If the monetary authorities
refuse to allow the exchange rate to depreciate, this rise
in the rate of interest will be pushed to the point at which
sufficient unemployment is caused to bring the rise in
money wages to an end. Thus, in an open system, even
the gradual rise in money wages which occurs before full
employment is reached is held under strong restraint, un-
less the movement is world- wide.

But even if the monetary authorities are not actuated by
the necessity of preserving stable exchanges they have
strong motives to prevent an immoderate rise of prices,
and legal restrictions upon the supply of money are
primarily designed to force them to do so. Moreover, a

1 See p. 150.

FULL EMPLOYMENT 19

lively sense 6f the horrors of inflation is sucked up with the
milk of the mothers of bank presidents, and, indeed, the
process of checking a rise in money wages is usually set to
work long before the point of full employment is actually
reached. 1 In the imagination of monetary reformers the
ideal banking system controls its policy in such a way as
to avoid unemployment. But in reality the first duty of
banking authorities imposed upon them alike by law,
tradition, and sentiment, is to prevent full employment
from being attained. And in the normal course they
carry out their duty with devotion and success.

It appears, then, that in normal circumstances the danger
of a complete collapse in the value of money is not greatly
to be feared, even when the level of unemployment is low.
First of all, the reaction upon expectations will usually
ensure that a rise in wages is at first very slow. Thus a
chance increase in the quantity of money, or in the mar-
ginal efficiency of capital, which is reversed soon after it
has occurred, will not precipitate an inflationary situation.
There is sufficient stability in the system to prevent it from
being pushed over the precipice at a touch. Second, a
persistent impulse to an increase in effective demand, even
if it is strong enough to carry the system to full employ-
ment, will be reversed before it has gone very far by a rise
in the rate of interest.

These safeguards are normally strong, but on certain
notable occasions in history they have failed to act. When
a movement towards full employment occurs in a situa-
tion in which the government cannot balance its budget,

1 When a precipitate rise in prices is once under way the task of the banking
system is likely to be rendered more difficult by a violent fall in liquidity preference.
When the expectation has once taken root that the purchasing power of money is
likely to fail in the future every holder of money has a powerful motive to get rid of
it, and in order to ensure a rise in the rate of interest the stock of money, measured
in wage units, must be reduced much below its former amount. Moreover, once
the expectation of a future rise in prices has increased the propensity to consume and
raised the marginal efficiency of capital, the rise in the rate of interest which will be
adequate to check investment and cause unemployment is very much increased.
It is for this reason that it is easier to nip inflation in the bud than to fell the bean-
stalk once it has begun to grow.

20 ESSAYS IN THE THEORY OF EMPLOYMENT

then the rope snaps which holds the value of money, and
the banking authorities can only gaze helpless over the
edge of the abyss.

The movement towards full employment may itself be
due to a heavy budgetary deficit, which leads to an increase
in effective demand. Or it may be associated with a collapse
in the exchange rate due to external causes. A violent
fall in the foreign exchange may lead to an increase in
home employment and set up, for this reason, a tendency
for money wages to rise. But it has aLo a more direct
effect. yThe sudden rise in the price of imported goods will
be followed by a rise in home prices and the cost of living
will be violently increased. Even apart from an increase
in employment this is likely to lead to a demand for higher
money wages, for the cost of living is a powerful influence,
though not the only influence, upon Trade Union policy. 1
A collapse of the exchange creates budgetary difficulties
where there are government obligations fixed in terms of
foreign currencies. The deficit may itself shake foreign
confidence and cause the collapse of the exchange. Or a
common cause, such as war-time activity, may be pro-
ducing both a deficit and full employment. Thus it is no
mere coincidence that a movement towards full employ-
ment, on several occasions, has occurred in conjunction
with a heavy budget deficit.

The existence of the deficit robs the monetary authorities
of their power. The government may be financing itself
by borrowing from the public, and, if so, they will not
countenance a determined effort to keep the rate of interest
high. But more often, in such cases, they are driven to
borrow from the central bank itself. Thus, since central
bank assets form the basis of credit, the authorities are

1 If the rise in prices is sufficiently severe the standard of life will fall (unless money
wages are raised faster than the exchange collapses) to the point at which physical
efficiency is impaired. A reduction in the number of unemployed workers will then
take place even if there is no increase in effective demand, for each employer, to
maintain his output, will have to employ more men to make up for the reduction in
output per man. This will reinforce the tendency for money wages to rise, and may
even lead to conditions of full employment. Cf. p. 133.

JOJLL EMPLOYMENT 21

Compelled willy-nilly to increase the supply of money, 1
and the rope frays and snaps in their hands. 2

The foregoing discussion suggests some reflections upon
the ideal policy for monetary authorities to pursue.
Obstacles, perhaps insuperable, to the control of employ-
ment and prices are presented by the fact that a regime of
private enterprise is subject to violent oscillations of senti-
ment, which must be counteracted by public policy if the
system is to run smoothly. 3 A discussion of these obstacles
lies outside the sphere of our present inquiry, but there are
other problems which it may be useful to discuss, even
upon the unrealistic assumption that these obstacles do
not exist. \,We will consider, therefore, a community with-
in which manipulation of the rate of interest, supplemented
by public-works policy, is adequate to control the level of
investment and consequently of employment. >

Let us first suppose that the object of policy is to main-
tain a high level of employment. In the real world, in
which labour is not perfectly mobile between trades and
localities, absolutely full employment will normally be
unobtainable. 5 And even if full employment were attain-
able, it would create, as we have seen, acute instability of
prices, a slight miscalculation in the forward direction
leading to a rapid and accelerating rise in money wages.
Our authorities must therefore be presumed to leave some

1 The popular view that a budget deficit is the cause of inflation can therefore
be justified. But it is unfortunate that this well-founded belief should have been so
commonly extended into the opinion that any public policy calculated to reduce
unemployment is 'inflationary' and should be avoided with care.

2 The most notorious of all inflations is not exactly fitted by this account, for the
Reichsbank in 1921 was acting upon a policy of 'meeting the needs of trade' and did
not attempt to limit the supply of money. The bank itself cut the rope. But even if it
had not done so, friction on the edge of the precipice would soon have worn it through.

8 General Theory, p. 320.

4 If a position is reached in which a level of investment adequate to promote the
desired level of employment is unattainable, it may be postulated that measures will
be taken calculated to increase the propensity to consume.

6 See below, p. 42.

22 ESSAYS IN THE THEORY OF EMPLOYMENT

margin of safety and to aim at a level of employment
somewhat short of full. 1

If Trade Unions are powerful this will entail that em-
ployment lies above the upper critical level, at which
money wages rise, and, supposing our authorities to obtain
their objective, prices will move constantly upwards,
though not in the violent manner characteristic of a very
close approximation to full employment. VThe policy of
maintaining a high level of employment will therefore
entail a cumulative increase in the quan/ity of money, at
whatever pace is sufficient to maintain the rate of interest
at the required level. If this policy is followed in one
country only, it will entail a cumulative fall in the foreign
exchange rate. For a country largely dependent upon
foreign trade changes in the outside world may, at fre-
quent intervals, place the desired level of employment
entirely outside the bounds of possibility. But the effects
of a rising level of wages, in itself, can be offset by a corres-
ponding depreciation of the exchange. 2 ()

rThe benefits of a high level of employment (supposing
it to be within the power of our authorities to obtain them)
can thus be secured at the expense of two evils, a constant
depreciation of wealth and incomes fixed in terms of
money, and the sacrifice of exchange stability. \ In so far
as the pursuit of such a policy tends to set up an expecta-
tion of rising prices in the future, this will, on our assump-
tion that movements of opinion never get out of hand,
lighten the task of the authorities in maintaining a high
level of employment, by increasing the propensity to
consume and increasing the inducement to invest. But
an expectation of a falling exchange rate in the future

1 A moderate amount of unemployment is not a very serious social evil so long as
it is distributed very widely over the working population and so long as dole provisions
are generous. It then amounts to an occasional enforced holiday on reduced pay,
which may be a hardship for some individuals but will be almost welcomed by others.

2 See below, p. 152. A movement in money wages is never likely to be so uniform
in its effect, as between industries, as a movement in the exchange rate. Relative
changes in employment arid in real wages are therefore likely to result from this
policy.

>ULL EMPLOYMENT 23

Creates serious difficulties, for it will set a premium upon
foreign as opposed to home lending and so tend to drive
up the home rate of interest. 1 Thus the policy of main-
taining a high level of employment can be pursued in one
country in isolation only if it is implemented by a high
degree of control over foreign lending. 2
( A more common objective of policy is stability of the
exchanges. This requires that so long as money wages
are rising the rate of interest must be continuously raised
in order to redrfcss the balance of payments. The level
of employment will therefore be forced continuously down-
wards until the point is -reached at which money wages
cease to rise. The policy required for exchange stability
xs a level of unemployment sufficiently high to prevent
money wages from rising. This is, in general, the basis of
the policy actually pursued by monetary authorities under
the regime of the gold standard. 3 )

A third objective of policy, which has been much can-
vassed, is stability of the price level. This conception raises
the question of what exactly is meant by the price level.
Stability of prices can only mean stability of a particular
index number, and when one index number is stable
others will be altering. In so far as stability of the price
level is regarded as desirable on grounds of social justice,
this objection is no mere quibble, but cuts at the root of
the whole idea. For the same policy which will ensure a
stable real income for one set of individuals will cause
fluctuations in the real income of others, who spend their

1 See below, p. 148.

2 The marked relative fall in the British rate of interest which has occurred since
the gold standard was abandoned in 1931 only became possible after expectations of
a further fall in the sterling exchange rate had disappeared. It was brought about
partly by a certain degree of official control over foreign issues, and partly by a
general political situation and uncertainty about the future course of world exchange
rates, which strongly discouraged foreign lending. In stable conditions a mere em-
bargo upon foreign issues would be insufficient to defend the home rate of interest
against outside influences, for it would always be open to holders of home securities
to transfer into foreign securities already in existence, or to lend at short term in
foreign money markets.

8 See below, p. 154, for a further discussion.

24 ESSAYS IN THE THEORY OF EMPLOYMENT

money on different commodities or on the same com-
modities in widely different proportions. 1 We will suppose,
however, that circumstances in our community are such
that this objection is unimportant and 'that an index
number can be chosen which gives reasonably consonant
results for all sections of the price level.

It is sometimes argued in favour of a policy designed to
stabilise prices that it will ensure stability of employment.
Now, it is true that a change in effective demand will lead
to a change in prices. But prices will also alter if either
efficiency or the level of money wages alters. ^Stable
employment will be associated with stable prices only if
efficiency is constant and employment lies within the
neutral range, so that the level of money wages is constant. 2
Any rise in money wages must be offset by a sufficient
reduction in effective demand to prevent prices from rising, 3
and an increase in effective demand will only be permis-
sible if wages fall. )

f If efficiency is increasing as time goes by, whether because
of improvements in technique or merely because of the
gradual accumulation of capital, then, with constant
money wages, there is a tendency for prices to fall. Stability
of prices then requires that the level of employment shall
be held sufficiently high to induce just that rate of rise in
money wages which will offset the effects of increasing
efficiency, and the level of employment must be made to
fluctuate with the rate of industrial progress. \

1 A reductio ad absurdum of the index number idea is provided by the use of the
Cost of Living Index, based on the consumption of a family with an income of 365.
a week in 1904, to regulate the bonus payable to First Class Civil Servants.

2 Cf. Harrod, The Trade Cycle, p. 117, where a similar argument is put forward.

8 The unreality of arguments conducted in terms of the price level is here clearly
exposed. When a general rise in money wages occurs it is inconceivable that the
authorities should be able to engineer a reduction in effective demand calculated to
restore the price of each particular commodity to its former level and the judgment
as to what is a sufficient reduction in effective demand to restore the average of prices
will turn. upon the index number which happens to be selected. The increase in
unemployment required to keep any given index number stable in face of a general
rise in wages will be greater the greater is the elasticity of supply of the commodities
entering into it. If the elasticities are high and the rise in wages large there may be no
decline in effective demand great enough to give the required result.

ffULL EMPLOYMENT 25

Thus a policy of maintaining stable prices (supposing
that such a policy can be formulated in a practicable
manner) is by no means equivalent to a policy of main-
taining stable employment.) Moreover, in so far as employ-
ment is the criterion, it is a high rather than a stable level
which is desirable. It can, indeed, be argued that a
moderate level of employment is the best objective, since
the attainment of a high level at one time may make a low
level harder to avoid at another, 1 but this is a question
which must be debated on its own merits. To introduce
the criterion of stable prices is merely to confuse the issue. 2

In so far as stable prices are regarded as desirable for
their own sake, as contributing to social justice, it must be
recognized that justice to the rentier can be achieved only
by means of the injustice to the rest of the community of
maintaining a lower level of effective demand than might
otherwise be achieved. We are here presented with a
conflict of interests which a priori reasoning can do nothing
to resolve. It is a conflict of which modern communities,
learning from experience rather than from the teaching
of economists, are becoming increasingly aware, and actual
policies are largely governed by the rival influences of the
interests involved.

( A fourth criterion of policy might be found in promoting
the best interests of wage earners. We are here presented
with a fundamental dilemma. ( An increase in employment
in the short period will normally be accompanied by a fall
in real wages, because, with fixed equipment, an increase
in output is accompanied by a rise in prices more than in
proportion to any rise there may be in money wages. 3

1 General Theory, p. 327.

2 The policy, which is sometimes advocated, of maintaining a constant level oi
money incomes is even more deleterious to employment than a policy of maintaining
prices stable. For it would require that employment should be kept below the upper
critical level even when there is an increase in efficiency. To examine the effects of a
policy of maintaining constant the 'effective quantity of money* (MV) would merely
be to explore the confusions which result from an application to actual problems of an
over-simplified Quantity Theory of Money.

8 In an open system it is possible that an increase in employment"may occur with-
out causing a fall in real wages. For if, when home investment increases, money wages

26 ESSAYS IN THE THEORY OF EMPLOYMENT

Thus we are presented with a choice between more em*
ployment with lower real wages and less employment
with higher real wages.)

It is important to observe that it is not always the same
individuals who gain on the one tack and lose on the other.
When effective demand increases, a man who is perfectly
secure of his job feels only the loss from the fall in real
wages, while a man who was formerly out of work feels
only the gain from the increase in employment. We must
therefore strike a balance between the interests of various
groups of workers. If unemployment was shared equally
amongst all individuals there would be, in any given
circumstances, one level of employment at which the
annual real income (from wages when in work and from
the dole during spells of unemployment) of a representative
worker would be maximised, and this level of employment
might be regarded as the proper objective of a policy
designed in the interests of labour.) But, in fact, unemploy-
ment falls far more heavily upon some individuals than
upon others, and there is no one level of employment
which can be regarded as the most desirable for all of them.>

It is clear, however, that weight must be allowed, upon
some system or other, to a fall in real wages as against an
increase in employment. In general it may be said that
something appreciably short of full employment must be
regarded as the optimum. In some circumstances there
will be a fairly clear indication of the most advantageous
policy, fAs employment increases from a low level the
corresponding fall in real wages will at first be very slight,
but after a certain point approximately full capacity will

rise and the exchange is held constant, the purchasing power of the wage earner
over imports may be increased to an extent sufficient to offset the diminution in pur-
chasing power over home produced goods. But, if a rise in money wages is not offset
by a fall in the exchange rate, not only will the proportion of secondary employment
falling abroad be increased, but the home rate of interest will have to be raised. The
increase in home employment which it is possible to engineer in these circumstances
will at best be extremely limited. (Cf. Kahn, 'The Relation of Home Investment to
Unemployment,' Economic Journal, June 1931, p. 175.) For the effect of an increase
in employment brought about by means of an increase in the foreign balance see
below, p. 161.

fee reached in a number of industries and a further inv
in employment will lead to a rapid rise in prices and ^
in real wages. 1 . In such circumstances it would be reason*
able to aim at a level of employment just short of the point
at which a sharp fall in real wages sets in.) Where no such
clear indication exists, and in particular, in an open system,
where the influence of the external position is important,
our policy can only be based upon a rough and indecisive
balance of contradictory considerations.

This argument applies to a strictly short-period point of
view. It is also necessary to balance the present against
the future. ( A high level of investment, though it may lead
to a low level of real wages in the present, is calculated to
produce a rising level of real wages in the future. ^ Thus
long-period considerations tell in favour of a higher level
of employment than that indicated by the immediate
effect upon real wages, and from a long-period point of
view the dilemma presented by the choice of an optimum
level of employment is much less acute. 2 J
vThe control of policy is, in a certain sense, divided
between the Trade Unions and the monetary authorities,
for, with given monetary conditions the level of the rate
of interest is largely determined by the level of money
wages. A sufficient rise in money wages will always lead
to a rise in the rate of interest and so check an increase in
employment. 3 Thus a combination of Trade Unions
sufficiently strong to control the level of money wages
would be faced with the problem of balancing a gain in
employment against a loss in real wages, and they might
be supposed to aim at establishing what appears to them
to be the optimum level of employment, taking account of

1 Cf. p. 130.

* An increase in employment brought about purely by an increase in the propensity
to consume has of course no beneficial long-period reaction upon the level of real
wages.

3 In arguing that a strong tendency for money wages to rise is in a sense inimical
:o employment, it is important not to lose sight of the fact that the existence of strong
Trade Unions, by reducing monopsony, tends to promote a high level of employment.
3f. p. 4, note.

2 8 ESSAYS IN THE THEORY OF EMPLOYMENT

the interests of employed and unemployed workers, anfl
allowing weight to future as against present benefits. The
two critical levels of employment at which wages alter
would then coincide at the optimum level of employment
and the optimum level would be endowed with all the
characteristics of full employment. 1 ^ In practice, as we have
argued. Trade Union policy is not conceived in these terms,
and even if it is possible to distinguish, in reality, a clearly
marked critical level of employment at which money wages
rise, there is no reason to suppose thatat corresponds to
any definite conception of an optimum level of employ-
ment.

Moreover, Trade Union policy is only very loosely co-
ordinated, and since the duty of each Union is to regard
only the interests of its own members, gains and losses are
very unevenly distributed between industries. Those
Unions which are in the strongest position (either because of
better organisation or because of a more favourable situa-
tion in their industries) will secure the greatest rise in money
wages when an upward movement occurs, and so secure
less than the average fall in real wages. And it is by no
means necessarily the case that those Unions which gain
the greatest real wages will suffer the largest share in un-
employment. f Trade Union policy, as it works out in prac-
tice, cannot be reduced to terms of even an unconscious
decision as to what is in the best interests of 'labour as a
whole 5 .

1 If the Trade Unions acting in concert were to decide upon a certain minimum
of real wages for employed workers, they would bring an increase in employment to
an end when that minimum was reached. Mr. Keynes' psychological interpretation
of full employment would then be the correct one, although the mechanism by which
the limit was set to employment would be a rise in money wages, not a direct 'with-
drawal of labour* (General Theory, p. 6). In stable conditions this policy would involve
a gradual increase in employment as capital accumulations and improvements in
technique raise the level of real wages corresponding to a given level of employment,
a higher rate of real wages in the present being secured at the expense, not only of
more unemployment in the present, but also of a slower rate of progress in the future.

MOBILITY OF LABOUR

IT is sometimes maintained that the causes of unemploy-
ment can be divided into two distinct groups: factors
which show themselves in a low level of effective demand,
and frictions wlrich prevent workers from moving from
place to place or from trade to trade in search of employ-
ment. One part of unemployment, it is said, can be cured
by increasing effective demand, while another part can be
cured only by removing the frictions. But it is doubtful
if this conception can be given any useful meaning, for
the frictions and the movements of effective demand act
and react upon each other a v nd are inextricably bound up
together.

On the one hand, the strength of the frictions largely
depends upon the state of effective demand. When the
general level of employment is high the frictions are weak.
Labour will be moved more easily from declining industries
or districts when other industries are expanding rapidly.
A man will move house or learn a new trade with a better
heart if he is certain of getting a job when he has done so.
A man who has thrown up one job will find another more
readily when the demand for labour is high. Employers
will find it less easy to indulge in the luxury of a 'reserve of
labour 5 when other firms are offering steady employment.
There will be an incentive to dovetail occupations when
workers are able to refuse to be lured into a seasonal
trade. The man who is 'unemployable' in bad times will
find, when labour is scarce, that his services are required
after all. Thus the frictions will be more easily and more
rapidly overcome the higher is the level of effective demand.

On the other hand, the frictions themselves modify the
action of the influences which lead to changes in effective

29

30 ESSAYS IN THE THEORY OF EMPLOYMENT

demand. The manner in which this occurs must V
examined at some length.

First of all, it is necessary to bear in mind the distinction
between specialised plant or land and specialised labour.
When plant or land is scarce relatively to the demand for
it, profit and rent incomes in the industries concerned are
raised relatively to wages, and thriftiness is consequently
increased. 1 It is therefore obvious in what way immobility
of these factors of production may exert a drag upon an
increase in effective demand. But the t influence of im-
mobility of labour cannot be so easily accounted for.

Before proceeding with our argument we must inquire
in what circumstances does the problem of immobility
arise. When workers are reluctant to move, they form
separate groups, each attached to a particular trade, and
the proportions in which the supply of labour is divided
between trades is determined by what has been the
demand for labour of each type in the more or less recent
past. If effective demand always moved up and down in
the same well-worn channels, a supply of each type of
labour would always be ready waiting to meet demand
for it, when effective demand expanded, and the question
of mobility would not arise. There would be no motive
for workers to move, and their willingness to move could
have no effect on employment. But actually tastes and
techniques alter as time goes by, while different causes of
expansion lead to demands for particular types of labour
in different proportions an expansion due to a fall in
the rate of interest will stimulate a different selection of
trades from those stimulated by an increase in the pro-
pensity to consume. Therefore any given increase in
effective demand is likely to reduce unemployment
amongst some groups of workers much more than amongst
others. This may react upon the level of effective demand
in three chief ways.
First, it much enhances the importance of the type of

1 See p. 10.

MOBILITY OF LABOUR 31

monopsony arising from an employer's convention against
bidding up wages when the demand for labour exceeds
the supply. 1 A scarcity of labour may be felt in particular
trades long before unemployment has everywhere dis-
appeared. Where the convention is strictly observed
wages fail to rise, but the price of the commodity rises to
the point at which the demand for it is equated to the
limited supply. Profits per unit of output increase, and
thriftiness consequently increases. 2 Thus increasing monop-
sony exerts a drag upon the increase in effective demand,
similar to the drag exercised by scarcity of specialised
plant.

We are here faced with a phenomenon which might,
perhaps, be correctly described as Trictional unemploy-
ment 5 . There are unfilled vacancies in some industries,
while there are unemployed men vainly hoping to find
work in others, and an increase in mobility, by making it
possible for employers in the trade where labour is scarce
to take on more workers at the ruling wage rate, would
lead to a net increase in employment. But the removal of
frictions, in this case, increases employment only because
it lifts the brake of increasing thriftiness which monopsony
applies to the wheel of effective demand.

Secondly, immobility of labour enhances the tendency
for money wages to rise (in the absence of monopsony)
when effective demand increases. Where labour is divided
into distinct groups there will be a critical level of em-
ployment, within each group, at which money wages begin
to rise, and this critical point is passed in some industries
when effective demand expands while unemployment in
other industries is still very great. Thus a rise in money
wages is likely to set in at an earlier stage when labour
is immobile than when it moves freely. A rise in money

1 Seep. ii.

2 If prices do not rise customers may be rationed by delay in meeting orders.
The customers are then induced to save while waiting to spend, and an increase in
thriftiness is brought about by a different route. Thriftiness is here used in the sense
of the propensity to save of the community as a whole; see p. 44, note.

32 ESSAYS IN THE THEORY OF EMPLOYMENT

wages imposes a check on the expansion of effective demand
by forcing up the rate of interest and reducing the balance
of trade. Moreover, a rise in wages which is not expected
to last has a powerful effect in reducing the incentive to
invest, 1 and it is likely that a rise in wages in particular
industries, which occurs when the general level of unem-
ployment is still considerable, will not be expected to last,
since -mobility is always greater in a longer run than in a
shorter run. Thus we find once more that immobility
constitutes an impediment to an expansion of effective
demand.

Thirdly, the existence of immobility will cause an ex-
pansion of effective demand to be accompanied by a change
in relative wage rates. If it should happen that labour of
the types required by consumption-good industries is in
general less plentiful than labour of the types required for
investment, an increase in effective demand is likely to be
accompanied by a relative rise in wages in consumption-
good industries. This limits the consumption of those
whose money incomes have not risen, including workers
in the investment industries. At the same time the mar-
ginal efficiency of capital is raised, and investment stimu-
lated. 2 Or, to look at the same thing in another way, a
rise in the price of labour in the finishing trades stimulates
the demand for its substitute labour devoted tp producing
capital instruments. Such an arbitrary alteration in
methods of production may be undesirable on general
grounds, but so far as the immediate effect upon employ-
ment is concerned it is likely to be beneficial on balance. 3

1 General Theory, p. 263. a Cf. p. 13,

8 Mr. Hicks (Economic Journal, June 1936, p. 245) suggests that the labour supply
available for consumption-good industries will often be fully absorbed into employ-
ment while there is considerable unemployment in the investment industries. But he
takes no account of the fact that a rise in relative money wages in the consumption
trades raises the marginal efficiency of capital, and so tends to keep the high rate of
investment alive, in spite of the rise in the rate of interest which may result from the
rise in money wages in the consumption trades. If, per impossibile, the rate of interest
was held constant when there was full employment in the consumption trades, the
rise of money wages in the consumption trades would ultimately lead to full employ-

1VSDBILITY OF LABOUR 33

There is, however, more reason to expect that, on
balance, wages in capital-good industries will be raised
relatively to wages in consumption-good industries. If
this occurs, the marginal efficiency of capital is lowered,
and the inducement to invest is reduced. Against this,
the consumption corresponding to a given level of employ-
ment in investment industries is increased. On balance,
the unfavourable effect upon employment is likely to
outweigh the favourable effect.

Lack of mobility into investment trades presents one
of the most formidable obstacles to an increase in employ-
ment when an increase in effective demand tends to
occur, and one of the most serious difficulties in the way
of a policy designed to maintain a high level of employ-
ment in face of a tendency for effective demand to decline.
If the capital-good industries cannot readily expand,
manipulation of the rate of interest will be powerless to
increase investment, and public works schemes will, beyond
a certain point, merely lead to a decline in private invest-
ment. Co-operation on the part of Trade Unions might
be conceived to prevent a rise of money wages from occur-
ring until the whole of the labour available in a particular
trade is absorbed into employment, but when this point is
reached further expansion depends entirely upon new
recruitment. A well-balanced policy of public works may
lead to an increase in employment over a wide selection
of capital-good industries, whereas concentration upon a
few lines will check investment at a lower level, when the
supply of one or two particular types of labour is fully
employed. If no more labour can be had for investment,
a further increase in employment in consumption-good
industries can come about only if the propensity to consume
is increased.

It is natural to expect that there should be a scarcity
of labour in the capital-good industries relatively to the
number which would have to be employed in order to
secure all-round full emnlovment. for. in the ordinarv

34 ESSAYS IN THE THEORY OF EMPLOYMENT

course of the ups and downs of trade, employment in the
capital-good industries must, in the nature of the case,
fluctuate more than the average of employment, 1 and it is
in the interests of the workers concerned to keep their
numbers down to minimum rather than maximum require-
ments. They are likely to be most successful in doing so
precisely in those skilled crafts for which substitutes are
hardest to find. Moreover, even if Trade Union regulations
are not framed with this object, workers will naturally
avoid, as far as they are able to choose, those trades in
which the chances of unemployment are most severe.
Thus workers will prefer to attach themselves to the
consumption-good industries, and the labour force of a
country may fail to divide itself between industries in such
proportions that full employment can ever be attained.
The only solution for this problem appears to lie in some
scheme by which the principal crafts in the investment
trades should have attached to them a number of reservists,
so that a man trained, say, as a bricklayer would normally
look for employment in some other line, but would be pre-
pared to assist the expansion of the building industry when
an impulse towards increased investment took place.
We have now seen that immobility may obstruct the

1 The marginal propensity to consume is less than the average; the Multiplier is
consequently less than the ratio of employment in the whole of industry to employ-
ment in the capital-good industries. Suppose that the Multiplier is 2, and that capital-
good industries represent of the whole. Then a decline of 20 per cent, in investment
will bring about a decline of only 4 per cent, in consumption-good industries. The
observed fact that capital-good industries are the most fluctuating is sometimes
accounted for in another way. A given proportionate change in the level of consump-
tion, if it is expected to last, will lead for a time to a larger proportionate increase in
the output of capital goods. Suppose that in the first position plant is exactly adapted
to output, and that 10 per cent, of plant is renewed every year. An increase of 10 per
cent, in consumption in the course of a year will require an addition to plant of the
order of 10 per cent, of the existing stock, which entails an addition to the rate of out-
put of capital goods of the order of 100 per cent, if the change in the stock of capital
is made within a year. There is no conflict between these two explanations. The
former deals with the effect of a primary change in investment upon consumption
and the latter with the secondary reaction upon investment of the increase in con-
sumption (cf. Harrod, The Trade Cycle, p. 53 et seq.) and the actual relative
fluctuations in investment and consumption are the result of both influences acting
together.

MOBILITY OF LABOUR 35

influences which cause effective demand to expand, by
Sallowing monopsony profits to increase, by causing the
general level of money wages to rise, or by causing a rela-
tive rise in wages in the capital-good industries. In a
certain sense, therefore, unemployment may be said to be
due to lack of mobility. But immobility in itself cannot be
regarded as a direct cause of unemployment. If we assume
that there is no monopsony, that the rate of interest is held
constant in face of rising money wages, that foreign trade
may be neglected, that entrepreneurs expect each week
that the present level of wages will continue to rule in the
future, and that on balance relative wages, as between
consumption- and capital-good industries, do not alter,
then we have assumed away all the channels through which
immobility can affect employment, and in such a state of
affairs no change in the degree of mobility would have any
effect upon the total level of employment. It is therefore
impossible to distinguish a part of unemployment which
can be attributed in any simple and direct way to the
existence of immobility alone.

We may now examine the question whether, when
mobility is low, unemployment can be caused by a change
in the relative demands for particular commodities, as
opposed to a fall in the general level of demand. The dis-
tinction between a change in relative demands, total
income remaining constant, and a change in the level
of effective demand, is a matter purely of analytical con-
venience and does not correspond exactly with any move-
ment that can occur in reality. On the one hand, a change
in effective demand will be accompanied by changes in
relative demands, for when the income of an individual
increases, the increment of his expenditure will not fall on
various commodities in the same proportions as the average
expenditure, and, when total income increases, the incomes

36 ESSAYS IN THE THEORY U* rMPLUYMENI

of individuals will increase in different proportions. On
the other hand, a change in relative demands may itself
constitute a change in the level of effective demand.
When the taste for some commodities is increased at the
expense of others, the distribution of income as between
wages and profits, and therefore the thriftiness of the
community, may be altered by the change. Moreover, a
change in relative prices may react directly upon thriftiness.
A fall in the relative price of motor cars may deflect
expenditure to them from other commodities, but it may
also cause them to be bought out of savings. A fall in the
relative price of bread may release expenditure for other
commodities, but it may also increase the margin of income
devoted to saving. The attraction of expenditure, as
opposed to saving, largely depends upon the composition
of the bundles of commodities .that can be bought for a
given sum.

If we wish to distinguish a purely relative change in
demand we must postulate that an increase in expenditure
upon one commodity is exactly offset by a reduction in
expenditure upon others, and that the increase in saving,
due to increased incomes from the one, is exactly offset
by reduced saving, due to diminished incomes from the
others. The latter condition requires that the elasticities of
supply of all the commodities are the same over the
relevant range, and that the proportion of wages to profits
is the same in all the industries, so that the shift in demand
leads to no net change in profits, and that the recipients
of profits are equally wealthy and have the same disposition
towards saving in all industries, so that, when profits are
unchanged on balance, saving out of profits is unaltered.
Further, our condition requires that saving out of wages
and dole provisions are everywhere alike, so that no change
in thriftiness results from a redistribution of employment
between industries; and that a change in real income, as
between persons engaged in different industries, does not
enhance or reduce the inequality of incomes sufficiently to

MOBILITY OF LABOUR 37

affect thriftiness. If these conditions are fulfilled, no net
change in employment would result from the shift in
demand. But it is unlikely that all these conditions should
ever be satisfied, or that the effects of a failure of one
condition should be exactly offset by the effects of an
opposite failure in others. A purely relative change in
demand is merely a convenient theoretical fiction.

In some circumstances these conditions will clearly fail
to be fulfilled. For instance, when expenditure is trans-
ferred from an industry where real incomes are relatively
low to where they are relatively high, thriftiness is increased.
Thus a transfer of demand to the products of manufacture
at the expense of the products of a peasant community is
likely to reduce employment. Again, if there is a transfer
of demand to a particular commodity whose supply is
limited by the plant available to produce it, then profits
will increase in the industry producing it by more than they
decline in the industries from which demand is transferred,
since the short-period elasticity of supply of a commodity
is smaller the more closely capacity output is approached.
The transfer of demand therefore causes an increase in
the ratio of profits to wages for the country as a whole,
and consequently increases thriftiness and causes unem-
ployment. Similarly, if a limitation upon the supply of a
particular commodity is set by the available labour, and
the convention against raising wages is upheld, then a
transfer of demand to this commodity will lead to an
increase in monopsony profits, and so increase thriftiness
and cause unemployment.

But there is no presumption that a rise in money wages
in a particular industry, in itself, will affect employment
in one direction rather than the other. Employment
increases by less in the expanding industry if money wages
rise, but in other industries it decreases by less, since rela-
tive prices in them are now lower, while the total of money
incomes is higher, and if the conditions set out above are
fulfilled the changes in activity in the two sets of industries

38 ESSAYS IN THE THEORY OF EMPLOYMENT

will exactly compensate each other. The level of employ
ment is then not directly affected by the rise in wages in
a particular industry. A rise in wages in the expanding
industry is more likely to occur the lower the degree of
mobility of labour, but, so far as the direct effect of changes
in relative wages is concerned, the degree of mobility does
not affect the probability that any given change in relative
demands will reduce employment.

The reaction of changes in relative demands upon
investment must also be considered. Any transfer of demand
between different commodities will set up a tendency for
investment to be made in equipment for the expanded
industries, and disinvestment in equipment for the con-
tracted industry. In general, investment takes place more
rapidly than disinvestment, since it is characteristic of
capital goods to require less time to construct than to
wear out. Thus the investment will take place at a higher
rate over a shorter time than the disinvestment. Employ-
ment will consequently increase at first, while the rate of
investment exceeds the rate of disinvestment, but disinvest-
ment will continue after the investment has come to an
end, and employment will then decline. Taking a suffi-
ciently long period into account the two effects may be
expected roughly to balance each other.

3

We may now consider the effects of a change in the
relative prosperity of different parts of the same country.
Let us suppose that there is an increase in investment
confined in the first instance to one centre, say Alpha,
which contains a representative sample of the population
and the industries of a country. As well as the immediate
increase in employment in investment industries, a large
part of the resulting secondary increase in employment
takes place inside Alpha, since expenditure on consump-
tion necessarily falls to a considerable extent upon locally

M)BILITY OF LABOUR 39

produced goods and services. The remainder is spread
thinly over the other centres, who find their exports to
Alpha increased and, since prices in Alpha are likely to
rise, their imports from it diminished. 1

The cost of living in Alpha has now risen and, if money
wages are constant, real wages will have fallen. The
inducement to labour to move is then altered. Relative
real wages in Alpha are reduced, but the chance of getting
a job in Alpha is greater than elsewhere. If workers were
influenced only *by relative real wages (other differences
between centres, in conditions of work, general amenities
and so forth, being unaltered) emigration from Alpha
would take place. But the notion of an increase in employ-
ment in one centre leading to emigration from it is obviously
absurd. It is natural to assume that workers are influenced
almost entirely by the chance of finding a job, and that
relative real wages exercise only a slight pull upon move-
ments of labour. Thus even if real wages in Alpha have
fallen immigration into Alpha may be expected to take
place.

It is to be observed, however, that even a small amount
of unemployment is a severe obstacle to mobility. Unem-
ployment in Alpha may have fallen to 5 per cent when it
is still 15 per cent in other centres, and any individual
who moves to Alpha will improve his chances of finding a
job. But with even 5 per cent of unemployment at home
the Alphans will look askance at immigrants, and the
uncertainty of finding a job will still be so great as to
deter many individuals, particularly married men, for
whom migration involves either moving house or incurring
the expense of a double establishment, from taking the
plunge. Thus immigration is likely to be on a relatively
small scale so long as appreciable unemployment remains
in Alpha, and to set in with a rush if unemployment falls
very low.

1 The manner in which Alpha's Balance of trade is financed is discussed below,
p. 149.

40 ESSAYS IN THE THEORY OF EMPLOYMENT

If money wages in Alpha rise, then a unit of labour
in Alpha will purchase less of local goods and more of
imported goods than before; therefore if imported goods
play a sufficient part in wage earners' consumption, the
level of real wages in Alpha may be regarded as greater
hlan it was before the initial increase in employment took
tpace. 1 Outside Alpha, the rise in price of imports from
Alpha will lower the level of real wages, though the effect
will be spread thinly over the whole country and will
not be great in any one centre. A rise- in relative real
wages in Alpha will increase the incentive to immigration.

Immigration will lead to a further increase in the level
of employment in Alpha. Unemployed workers must make
some expenditure, and part of their expenditure will fall
on local goods wherever they may be at the moment.
Immigration will therefore lead to a further increase of
effective demand in Alpha, and a reduction of effective
demand in other centres. Even if it is the rule that the
unemployed are supported by a levy on the inhabitants
of the centre where they happen to be at any moment
an influx of workers into a centre is likely to increase
expenditure there, because it will cause a transfer to them
of income from wealthier inhabitants of the centre. 2
In other centres both employment and unemployment
will decline as the result of emigration, Moreover, if the
alteration in the geographical distribution of the popula-
tion is expected to be permanent, there will be a secondary
inducement to increased investment in Alpha, particularly
in house-building, accompanied by disinvestment in other
centres. Thus any initial difference between the levels of
activity in different centres will tend to become cumulative.

Let us now consider another aspect of the problem.
Suppose that Alpha suffers from a relative decline in
effective demand. Clearly if the fall in effective demand in

1 That is to say, the favourable effect upon the terms of trade of raising money
wages may outbalance the tendency for increased home employment to lower real
wages. Cf. p. 25, note, and p. 163.

Of. p. 47.

MUBIL1TY OF LABOUR 41

Alpha is accompanied by an increase in effective demand
in other centres, then the extent of unemployment amongst
Alpha workers will depend upon their willingness to
migrate to the centres where prosperity has increased.
Unemployment in Alpha is then rightly attributed to the
'difficulty of transfer 5 and the proper remedy, from Alpha's
point of view, lies in schemes designed to increase mobility
of labour.

But it may be that the increase in demand is confined
to a remote centre, say Omega, to which emigration is out
of the question. Alpha has become a depressed area, and
low profits and unemployment in Alpha will continue
until something happens to increase effective demand in
Alpha sufficiently to compensate for the initial decline.
The increase in prosperity in Omega is no manner of
consolation to the inhabitants of Alpha, and the trouble,
from Alpha's point of view, is something worse than the
transfer problem.

At the same time, if the neighbouring centres of Beta
and Gamma are prosperous, emigration to them will be
of some assistance to the Alphans. And if a direct attack
upon the problem by increasing investment in Alpha itself
is difficult to bring about, an increase of investment in
Beta and Gamma will be the next best thing. 1

4

No precise meaning can be given to the notion of full
employment for a system within which mobility of labour
is imperfect, unless it is taken to imply a state of affairs in
which there is complete full employment all round, that
is to say, in which no available labour is unemployed in
any district or in any occupation. It is commonly said

1 The problem of the British export industries at the present time [1936] is partly
the result of the general decline in effective demand, of which a large share has fallen
to their lot. In so far as it is a special problem, it may be illustrated by the above
argument if we read for Alpha, say, Lancashire, for Omega, the Far East, and for Beta
and Gamma, the south of England.

42 ESSAYS IN THE THEORY OF EMPLOYMENT

that there is TulP employment when the only unemploy-
ment which exists can be described as frictional. 1 But it is
impossible, as we have argued, to give precision to the
concept of specifically 'frictionaP unemployment, because
it is impossible to make a hard-and-fast distinction between
unemployment which is due to frictions and unemploy-
ment which is due to a deficiency of effective demand.

So long as frictions are strong and demands are con-
stantly changing, absolute full employment, as we have
seen, is unlikely to be obtainable, for at any particular
moment full employment for all labour would involve a
demand for particular amounts of particular types of labour
in particular places, and, if precisely those types are not
available, some types will be fully employed while other
types are available but arc not required. It seems prefer-
able to say that full employment, in a precise sense, can
never be attained so long as frictions exist, rather than to
use 'full employment 5 in an imprecise sense in which it
can be said to be attainable, such unemployment as remains
being vaguely attributed to frictions.

One of the most striking departures which Mr. Keynes
has made from traditional economic teaching is his demon-
stration that an increase in thriftiness will not, in general,
tend to increase the rate at which capital is accumulated
except in so far as the effect of a decline in activity in
reducing the rate of interest outweighs its effect in
reducing the marginal efficiency of capital. The classical
doctrine, that an increase in thriftiness will lead to an
increase in accumulation equal to the increase in saving
that would take place if activity did not decline, is true
only if full employment is always maintained. If full

1 Cf. General Theory, p. 6. Here Mr. Keynes is advancing objections to the attempt
to apply the classical system of analysis, which postulates that there is in some sense
full employment, to the real world, and in this context a definition of frictional un-
employment is necessary. If we believe that in reality employment tends towards a
particular level, then there must be in reality a particular level for employment to
tend towards. But if we regard full employment simply as the final upper limit
beyond which employment can never go, it is possible to recognise that approxima-
tions to it achieved in reality are merely a matter of degree.

M'OBILITY OF LABOUR 43

employment is guaranteed, then, when thriftiness increases,
a corresponding increase in investment must be brought
about in order to fulfil the guarantee that is to say, an
adequate fall in the rate of interest must be enforced by
the authorities, or brought about by a sufficient reduction
in the level of money wages ^ But if labour is immobile
between industries it becomes impossible for the guarantee
to be fulfilled. An increase in thriftiness releases from the
consumption-good industries labour of particular types
which is useless to the capital-good industries. An upper
limit is set to investment by the labour already available
to the investment-good industries, and it is impossible to
induce the expansion in investment which would remove
the unemployment that has been created by the increase
in thriftiness. Moreover, ince unemployment has now
appeared in the consumption-good industries, wages in
those industries are likely to fall. This fall in relative
wages will strongly reinforce the effect of increased thrifti-
ness in reducing the marginal efficiency of capital and
considerably outweigh the beneficial reaction upon the
rate of interest that accompanies it. Thus, when labour is
immobile, the notion that a fall in money wages may be
relied upon to stimulate investment when thriftiness
increases is deprived of all plausibility, and intervention
by the authorities is necessary to maintain even the pre-
existing level of investment.

1 In practice employment cannot move beyond certain wide limits. Unemploy-
ment cannot fall below zero, and cannot rise to a very high level, for, when it does,
the situation becomes so intolerable that some action is taken to remedy it. Thus in
respect to very large changes in thriftiness the postulates of the classical analysis are
very roughly fulfilled. If we were to imagine a Nineteenth Century that had con-
ducted itself according to the precepts of Mr. Keynes, so that thriftiness was actively
discouraged, while full employment was always maintained, and contrast it with the
Nineteenth Century which received the blessing of the classical economists, there can
be little doubt that, although the level of employment would have been higher, the
accumulation of capital would have been very much less, in Mr. Keynes' Nineteenth
Century than in the Nineteenth Century which actually came to pass.

CERTAIN PROPOSED REMEDIES FOR
UNEMPLOYMENT

IN a community with given resources and methods of
production the amount of output can only be increased
by means of an increase in the rate of investment or of a
reduction in the propensity to save, 1 and any scheme which
is suggested as a method of increasing employment must be
examined in the light of its effect upon these influences.
Many schemes have been suggested in recent years. A
theoretical examination is here made of four of them:
(i) a compulsory reduction in the length of the working
day, (2) a rise in the school leaving age, (3) emigration,
(4) a subsidy to wages.

Each of these schemes has been suggested as a remedy
for unemployment from the point of view of a closed
system, and the effect of the schemes upon foreign trade
may be neglected in our present inquiry. 2 Further, the
schemes will be treated as though they affected all indus-
tries within the country in the same way. This is a severe
simplification of the actual problem. Obviously the effect
upon a particular industry of a compulsory limitation of
hours to a certain maximum will vary according to the
difference which is made to its actual hours by the change,
according to the shift system which it was formerly using,
and according to other technical conditions. A rise in
the school leaving age will have a greater effect upon
retail trading than upon the production of iron and steel.

1 The propensity to save or thriftiness of the community represents the relationship
between the level of total real income and the rate of saving in real terms. It embodies
exactly the same relationship as Mr. Keynes' 'propensity to consume' (General Theory,
p. 90).

2 We shall find that some of the schemes which we have to discuss are calculated
to raise money costs of production in the country which introduces them, and so to
reduce its foreign balance. Others, by lowering home prices, are calculated to increase
it. In any actual case reactions upon the foreign balance would be of the utmost
importance in assessing the merits of the schemes.

44

CERTAIN PROKjabJJ KUMEDIES FOR UNEMPLOYMENT 45

The introduction of either scheme would affect the costs
of various commodities relatively to each other, and some
change in net employment might result from the alteration
in the relative positions of different industries. But it is
legitimate to ignore these considerations in a theoretical
examination of the proposed schemes, since their merits are
not regarded as in any way depending on relative changes.
It is also legitimate to assume that the schemes are faith-
fully carried out, so that an industry with unscrupulous
employers and ^veak Trade Union organisation cannot
steal a march upon law-abiding rivals, and that each
scheme, when it is introduced, is expected to be permanent.
Finally, it will be assumed that any monetary repercussions
of the schemes x are offset by appropriate action by the
banking system and that the rate of interest is unaffected
by them, for if the object in view were merely to produce
an effect on employment by changing the rate of interest,
simpler methods could be used. These qualifications may
be, of course, of great practical importance in any par-
ticular case. The justification for introducing them lies
in the fact that these qualifications leave intact the premises
upon which the simple arguments in favour of the schemes
have been based. If advocates of the schemes fall back
upon more sophisticated arguments which these qualifica-
tions would exclude, then the case must be re-examined
accordingly.

The scheme for a compulsory reduction in hours of work
has been proposed in two forms, with a constant hourly
or piece rate of money wages, and with constant weekly
money earnings. These two proposals appear at first sight
to be very different, but from the point of view of our
present inquiry the difference is less than might be sup-
posed. For \ve have ruled out of account both changes in
the foreign balance and changes in the rate of interest,
which represent the most important influences exerted

46 ESSAYS IN THE THEORY OF EMPLOYMENT

by a change in money wages upon the level of employ-
ment. It may be further assumed that entrepreneurs
(including shareholders) and rentiers are alike in their
saving propensities, so that the redistributional effects of
a change in the value of money have no reaction upon
thriftiness, 1 and that, since the change in wages is expected
to be permanent, the marginal efficiency of capital is un-
affected. 2 It follows that, in the case that we have to
consider, a change in money wages, in itself, produces no
effect upon employment or upon real wjages. 3 Thus, for
the purpose of our present inquiry, it makes no difference
whether the reduction in hours is accompanied by a rise
in money wages per hour or not. We may therefore con-
fine our argument to the case in which hourly wages are
unaltered.

A reduction in the length of the working day is likely
to lead to some increase in hourly output, because of the
reduced fatigue of workers; on the other hand some opera-
tions, such as cleaning or warming up the plant, or walking
from the shaft to the coal face, have to be undertaken in
respect to each day's output, no matter what the length of
the day. Thus, with constant hourly wages, labour cost
per unit of output may on balance be either increased or
diminished. If it is diminished, prices will fall and real
wages per hour will increase, while the increase in numbers
required to produce a given output will be less than in
proportion to the reduction in hours; and conversely if
labour cost is raised. In any case there must be some in-
crease in the number of workers employed to produce a
given output. 4 From this an important consequence follows.

1 General Theory, p. 262. 2 Ibid., p. 263.

8 It is also necessary to assume that the new legislation does not alter the bargain-
ing position of workers vis-d-vis employers, so that the degree of exploitation is
unaffected.

4 It is well known that employers may be so foolish as to set the working day at
a length greater than that which gives the maximum daily output per man. In such
a case a reduction of hours is in the interests both of employers and of employed workers,
and ought to be made on its own merits. In discussing a proposed remedy for un-
employment it is natural to take our departure from a situation in which a reduction
in hours will lead to a reduction in daily output per man.

CERTAIN PRUFUSHD REMEDIES FOR UNEMPLOYMENT 47

The number of workers who are unemployed, when a
given output is being produced, is reduced by the
reduction in hours and consequently relief payments are
reduced. The dole (which may be taken to stand for all
expenditure by the unemployed) may be financed out of
borrowing by the authorities, out of dis-saving by the un-
employed, out of charity or out of taxes. In so far as it
comes from charity or from taxation, a reduction in dole
payments may lead to an increase in expenditure by
charitable persons or taxpayers, but it is natural to suppose
that out of the total increase in the net income of these
persons a part is devoted to saving. And in so far as the
dole is financed by borrowing, a reduction in dole pay-
ments will reduce dis-saving by the state. ^Thus a reduction
in hours, by reducing the amount of dole payments
corresponding to a given output, brings about an increase
in thriftiness. The magnitude of the effect will vary with
the manner in which provision is made for dole payments,
but the direction of the effect will be the same in each case.
In discussing this and the following schemes we will treat
a reduction in the amount of dole payments corresponding
to a given level of output as constituting an increase in
thriftiness, without pausing to inquire in each case how
the dole is provided for. 1 ^

It can now be seen, assuming for the moment that
investment is constant, that the reduction in hours must
lead to a decline in output. For if output were constant
total saving would be increased. Therefore if output were
constant consumption would decline. It follows that out-
put cannot be constant, but must decline. 2 The effect of

1 Against the reduction in dole payments must be set the fact that the workers
formerly employed suffer a decline in daily real earnings, when hours are reduced,
which may decrease their savings by more than the savings of newly employed workers
are increased. This effect, however, is likely to be slight compared to the increase in
saving due to the reduction in dole payments, and it will be left out of account both
in discussing this scheme and the others which follow.

2 If investment is constant, output must fall to the point at which, with the increased
propensity to save, the total of saving is no greater than before. In the limiting case,
where no decline in saving accompanies a decline in output except that due to in-
creased dole payments, the total of dole payments cannot alter. The number of men

48 ESSAYS IN THE THEORY OF EMPLOYMENT

the shortening of hours is therefore to reduce the total
of employment (measured in man-hours) while increasing
the number of workers employed.

We must now consider the reaction of the scheme on
the rate of investment. A reduction in consumption has,
in itself, a tendency to reduce investment, for the capital
equipment required by a lower level of output is less than
for a higher level. On the other hand, a reduction of
working hours, unless it is accompanied by an increase in
the number of shifts per day, reduces the effective amount
of capital. The hours of work of a machine are limited by
the hours of labour, and a machine working for six hours a
day represents a smaller amount of productive capacity
than the same machine working for eight hours. A given
output would therefore require more equipment after a
reduction of hours, and the marginal efficiency of capital
would be raised by the, change. 1 This tends to bring
about an increase in investment (or reduction in the rate at
which plant is being allowed to fall into decay). This
increase in investment is of the same nature as that which
can be induced by setting fire to a number of factories,
or any such method of reducing the stock of capital.

According as investment, on balance, is increased or
reduced, the effects of increased thriftiness are mitigated
or enhanced. But changes in the rate of investment, in
either direction, will persist only for a certain time, for
there will be a certain adjustment of the stock of capital
which it is appropriate to make, and when it has been
made the effect of the change in hours upon investment

unemployed cannot then be reduced, and the total of man-hours of employment
must be reduced in the same proportion as hours per man. Normally saving falls off
with output for other reasons besides the increase in dole payments (General Theory,
p. 251), and some reduction in the number of men unemployed must therefore
occur.

1 A reduction in hours of work also increases the cost of equipment per unit of
output, and this effect tells in the opposite direction. If the elasticity of substitution
between capital and labour was sufficiently great there would be a reduction in the
amount of equipment employed to produce a given output and consequently a
reduction .in the inducement to invest. But in general the effect of the reduced
quantity of capital is likely to be the predominant one.

CERTAIN PROPOSED REMEDIES FOR UNEMPLOYMENT 49

will have worn itself out; while the effect of increased
thriftiness will be permanent.

A reduction in hours of work may be desired for its own
sake. The decision to sacrifice real earnings for the sake of
a reduction in hours of work has to be made collectively,
since in modern industrial conditions it is not possible for
an individual to vary his hours and his earnings according
to his particular fancy, and if a situation has been reached
in which the workers as a whole prefer to forgo a part of
their real earnings in exchange for an increase in leisure,
it may be considered desirable that their decision should be
implemented by a general agreement to reduce hours of
work. But at the present time [1936] a reduction of hours
is advocated, not because it is held that the level of real
wages is so high that leisure is to be preferred to goods,
but because it is regarded a t s a remedy for unemployment.
How does the existence of unemployment affect the merits
of the case for shorter hours? On the one hand the desire
of the individual worker for leisure is likely to be weakened
by the prevalence of unemployment. It is true that real
wage rates normally rise as employment falls off, for with
given equipment and organisation rising marginal cost
normally prevails and prices fall more than money wages as
output declines. But for the representative worker the
average of real income, in and out of work, over a period of
time is likely to be less the lower the general level of
employment and the more frequent his own spells of un-
employment. And with a lower real income the preference
for leisure over goods is reduced. On the other hand,
taking one worker with ahother, the sacrifice of real income
due to shortening hours is much less when unemployment
prevails, for the reduction in hours of work for those already
employed is partly offset by a reduction in the numbers of
those who are doing no work at all. In conditions of full
employment a reduction in hours would reduce total
output in almost the same proportion as hours, while when
unemplovment prevails the reduction in output will be in

50 ESSAYS IN THE THEORY OF EMPLOYMENT

a smaller proportion. 1 For the workers as a whole leisure
is cheaper when unemployment is high.

A reduction of hours would certainly be desirable if no
other method of providing for the unemployed were feasible
in the political situation of a particular community. And
it will clearly be favoured by taxpayers upon whom the
duty of providing dole payments would otherwise fall.

It may be held that a redistribution of income between
the employed and the unemployed is desirable, and that a
given amount of employment should be spread over as
many individuals as possible, so as to reduce the suffering
and the deterioration resulting from long spells of un-
employment. 2 This, indeed, provides the main argument
in favour of the scheme, but, by its very nature, this
argument proclaims the scheme to be a counsel of despair.
A reduction of hours is not in any sense a remedy for a low
level of employment. It merely leads to a redistribution
of a smaller total of employment amongst a larger number
of individuals.

The first effect of raising the school leaving age is to
remove from the labour market a number of children
who would otherwise have joined the ranks either of the
employed or of the unemployed. For output to be main-
tained, jobs vacated by children must be filled by older
workers. It has been argued 3 that, because older workers
are customarily paid a higher rate of wages so that the costs
of industry will be raised, therefore output will decline.
This argument is fallacious, and must be set against the
contrary fallacy that exactly the same cause, by raising
purchasing power, will lead to an increase of output. The

1 Except in the limiting case discussed above, p. 47, note.

2 This argument has the greatest force if particular workers fall out of employment
and are not re-employed until effective demand recovers, so that the 'unemployed'
and the 'employed' are two distinct sets qf individuals. The best of both worlds could
be had by a scheme under which when, say, a quarter of the workers are unemployed
each man works for three weeks in a month and draws a dole in the fourth week.

3 For instance, by Lord Hugh Cecil in a letter to The Times, 3rd August, 1935.

CERTAIN PROPOSED REMEDIES FOR U N EMPLOY ME JN 1 51

effect of exchanging higher for lower paid workers will lead
to a rise of prices and leave output unaffected (for we are
neglecting reactions upon foreign trade and upon the
rate of interest, and the effect of impoverishing the
fixed income classes) . The real wages of those already in
employment will fall, but the average level of real wages
will be unaltered. 1

Some alteration in methods of production may follow
from the change. One man with a van might be employed
in place of six errand boys. A period of increased invest-
ment might set in while alterations in equipment were
being made. But this effect is likely to be slight, for
workers are not divided into two classes, children and
adults. The intermediate classes of young workers between
the ages of say, fifteen and twenty-one, provide close
substitutes for the children^ retained at school. There may
also be some investment in school buildings to accom-
modate additional pupils. At most, any increase in the
rate of investment due to these causes would endure only
a short time, and must be left out of account when the
permanent effects of the scheme are being considered.

As before, the permanent effect of the scheme is to reduce
output. For if output were to remain unchanged a number
of older workers would be employed to compensate for
children who would have been employed if they had left
school. The amount of dole payments would then be
reduced. Against this must be set any maintenance
allowances which are paid in respect to school children, 2
and any expenditure made on their behalf which depletes
the savings of their parents. In general it may be supposed
that these items are small, and that the reduction in saving

1 If the older workers are more efficient than the school children the rise in prices
will be less than in proportion to the rise in average wage rates, and if total output
were unchanged real wages would be raised, while the number of older workers
finding jobs would be less than the number of children removed from employment.

2 Provided that the allowances are financed in a manner which reduces thriftiness
to the same extent as an equal amount of dole payments. If the dole is financed by
borrowing and allowances by taxation, allowances are less than equivalent to dole
payments in their effect upon thriftiness.

52 ESSAYS IN THE THEORY OF EMPLOYMENT

made in respect to an older worker receiving the dole
is greater than in respect to a child remaining at school.
It follows that expenditure would not remain constant
if output were unchanged, and that output consequently
cannot remain unchanged, but must fall. As before, a
reduction in employment due to increased thriftiness will
tend to lead, temporarily, to a further reduction due to
reduced investment. The increase in thriftiness in this case
is much less than in the case of shortening hours, for the
closest substitutes for the school children will be young
workers, who are paid a lower rate of benefit than adult
workers. The reduction in dole payments brought about
by absorbing them into employment will therefore be less
than the reduction when a representative sample of workers
cease to receive benefit, as will occur when hours are
reduced. 1

In spite of the reduction in employment there will be a
reduction in the numbers of the unemployed. For it is
necessary to distinguish between an increase in employ-
ment and a reduction in unemployment. The amount of
unemployment is not co-extensive with the total of man-
hours which are spent otherwise than in gainful occupa-
tions. A man is unemployed when he is standing in the
street, or digging his allotment, if he would prefer to accept
some other occupation at the current rate of pay. He is
not unemployed if he is living in idleness upon the pro-
ceeds of accumulated wealth, or upon beach-combings,
provided that he has no desire to take service in organised
industry. The decision as to what is unemployment and
what is leisure is determined partly by the wishes of the
individual and partly by the collective choice of the

1 Let us assume that half of the children retained at school would have found work
if they had left, that the dole for a child is a quarter of that for an adult, and for a
young person half, and that output per head is unchanged. Then in the limiting
case, where the total of dole payments is constant (see p. 47, note), for every four
children retained at school two young persons will be employed and one adult unem-
ployed, provided that there is no dis-saving in respect to a child at school. If dis-
saving in respect to a child at school is equal to the dole of an unemployed child,
then one adult will be unemployed for eight children retained at school.

CERTAIN PROPOSED REMEDIES FOR UNEMPLOYMENT 53

community. If the standard length of the working day
has been set by Trade Union bargaining at eight hours, an
individual is not regarded as being unemployed if he is not
at work for the remaining sixteen hours of the day, even
though he would personally have preferred to work
longer at the ruling hourly rate. Similarly a child is not
regarded as being unemployed so long as he is suffering
whatever period of education society lias decided to impose
upon him, even though he would himself prefer to be
earning wages.

To increase the length of the education period is arbi-
trarily to remove a certain number of individuals from the
category of unemployment to the category of leisure. A
child who would have been unable to find work when he
left school is prevented from becoming unemployed, and
when a vacancy, which would otherwise have been filled
by a child, is filled by an older worker who would other-
wise have been unemployed, a year of unemployment is
exchanged for a year of education. Raising the school-
leaving age, in short, reduces the number of the unem-
ployed, in spite of the fact that it is likely also to reduce
the number of the employed. 1 The harmful effect upon
employment could be altogether eliminated by increasing
the rate of unemployment or maintenance allowances,
but it is to be supposed that so obvious a method of increas-
ing employment has already been rejected by a community
which is contemplating the more complicated expedients
that are here under discussion.

The question of whether it is desirable to lengthen the
period of education must be considered on its own merits,
and it might obviously be regarded as desirable even if
there was no unemployment to be considered. Even if full
employment obtained it might be considered advantageous
to reduce the supply of labour, and the level of real income

1 In the first example given on p. 52, note, there is a reduction in the number un-
employed of three individuals, and in the second example of seven individuals,
corresponding to a reductipn in the number of employed of one individual.

54 ESSAYS IN THE THEORY OF EMPLOYMENT

represented by the maximum output of industry, in order
to enjoy the benefit of improved education. In a community
which suffers from chronic unemployment the case in
favour of raising the school leaving age is greatly strength-
ened. From the point of view of the individual parent
the desire for education, like the desire for leisure, is likely
to be weakened, rather than strengthened, by the preva-
lence of unemployment. But, taking one family with
another, the sacrifice of wealth due to raising the school
age, which may be considerable if full employment obtains,
is almost nothing if unemployment is usually great.
Moreover, by its means enforced idleness can be turned to
good account, so that if unemployment were always
completely abolished by this means a low level of employ-
ment might turn out to be a blessing in disguise. Those
who place a higher value upon the advantages of education
than does the general run of public opinion, might take the
opportunity of a high level of unemployment to introduce
a reform which will persist into a period of high employ-
ment. But it must be regarded as a method of mitigating
the evil effects of a low level of employment, not as remov-
ing the cause.

Schemes of emigration will have important incidental

effects varying with the method by which a scheme is

financed, with the effect upon the age composition of the

population which remains behind, and with the connec-

ions preserved by the emigrants with the home country.

n order to reduce the problem to its bare essence we will

uppose that a number of families of average composition,

^ithout property, disappear from the country and are

Aever heard of again. 1 It is, indeed, somewhat absurd to

discuss the question of emigration from the point of view

of a closed system. Emigration may have important re-

1 The effects upon employment of mobility of labour within a community, from
the point of view of the whole community, are discussed above, p. 40. Some problems
connected with changes in population are discussed below, p. 95 and p. 116.

CERTAIN PROPOSED REMEDIES FOR UNEMPLOYMENT 55

actions upon the trade balance of the home country. If
foodstuffs are largely imported, emigration will reduce the
volume of imports corresponding to a given level of home
employment; while an increase in the population of the
outside world will tend to raise the demand for exports
from the home country. Such considerations may be of
preponderant importance in any actual case, and rriay
provide strong arguments in favour of emigration, but at
present we are concerned only with the view that emigra-
tion reduces unemployment by the simple process of
removing the unemployed, and in what follows we will
continue to abstract from reactions upon international
trade.

The first effect of emigration is to reduce dole payments
and so to reduce the level of employment. But it is clear
that employment cannot, in the first instance, be so much
reduced that unemployment remains the same. For, if
unemployment were unchanged, dole payments would be
constant, while output had declined, and saving would be
reduced. 1 Thus both employment and unemployment are
reduced in the first instance.

The decline in consumption consequent upon emigra-
tion will reduce the demand for capital goods, including

1 In the limiting case (p. 47, note) dole payments cannot alter and the amount of
unemployment is not reduced by emigration if the rate of dole per man is unchanged.
If the rate of dole payments were raised in proportion to the initial reduction in
numbers requiring relief, employment would be unchanged, and the number of the
unemployed would be reduced by the number of emigrants. But, once more, we must
suppose that the community we are considering has some strong objection to so obvious
a remedy for unemployment. For instance, it was argued in the report of the Un-
employment Insurance Statutory Committee, 27th February, 1936, that a rise in rates
of benefit would be objectionable because it would cause some unemployed men to
receive a larger income than some employed men.

Under a scheme by which the unemployment fund is self-supporting the total of
dole payments must fall with the numbers of employed workers (in respect to whom
contributions are paid) . In such a case, if there were no change in saving when out-
put alters, apart from changes in the dole, complete instability would result, and any
fall in employment would become cumulative, up to the point at which the attempt
to make the fund self-supporting was abandoned. Normally saving falls off with
employment for other reasons besides an increase in dole payments, but the attempt
to make the unemployment fund self-supporting much exaggerates the movements of
employment, in both directions.

56 ESSAYS IN THE THEORY OF EMFLUYMEJN 1

dwelling houses, and a decline in the rate of investment
will set in. A further reduction in employment will there-
fore occur, and unemployment may become as great or
even greater than before. If a once-and-for-all emigration
takes place, a decline in the stock of capital, below what it
would otherwise have been, will be brought about after a
certain time has elapsed, and, if no other change in the
situation occurs, unemployment will gradually fall, but
the emigration will leave a legacy of permanently reduced
employment as a result of the increase 'in thriftiness due
to the decline in numbers. A constant stream of emigra-
tion will impose a constant check upon the rate of invest-
ment, and there will be no tendency for employment to
recover as time goes by.

While emigration may reduce the amount of unemploy-
ment in the home country, its effect in reducing employ-
ment is likely to be the most serious of any of the schemes
which we have discussed.

The scheme of a subsidy to wages has been proposed
in various forms. Two types of subsidy are here discussed.
Under the first, each entrepreneur is paid by the state a
certain proportion of the wage received by each man in
his employ, the sums required being recovered out of
additional income tax. It is convenient to assume that
money wage rates are unaltered.

From the point of view of the employer, the subsidy
has the same effect as an equivalent reduction in money
wage rates. Under our present assumptions, that all
industries are equally affected, that the scheme is expected
to last, that entrepreneurs are similar to rentiers in respect
to thriftiness, that reactions upon foreign trade may be
left out of account, and that the rate of interest is un-
altered, a reduction in money wages has no effect upon

CERTAIN PROPOSED REMEDIES FOR UNEMPLOYMENT 57

employment, 1 but merely leads to a proportional fall in
prices. 2

In so far as the scheme is equivalent to a reduction in
money wages, it is of no benefit to employment. But under
the scheme the money incomes of workers are unchanged
and real wages are raised by the fall in prices, while the
burden of taxation is increased. Thus, superimposed upon
the effect of a fall in money wage rates is the effect of an
increase in taxation the proceeds of which are paid to the
workers. The transfer of income from taxpayers to wage
earners has the effect of reducing thriftiness, since in
general taxpayers are more wealthy, and more incline4 to
save, than wage earners, and the decline in thriftiness
will lead to an increase in employment. 3 Thus a subsidy
to wages leads to an increase in employment in the same
way and to the same extent as any commensurate scheme
for taxing the rich and using the proceeds for the benefit
of the workers.

Under the second type of subsidy, the dole received by
a man when he is out of work is paid to his employer
when* he is given a job. Under this scheme there is no
increase in taxation, but the total of dole payments, after
the date at which the scheme is introduced, cannot fall
when unemployment declines.

On our assumptions, the scheme leads to no immediate
change in the inducement to invest, and we may assume
for the moment that employment in the investment
industries is unaffected by it. But thriftiness will be
reduced. Each employer receives the subsidy in respect
to any men in excess of the number employed by him at

1 General Theory, chap. 19.

8 It often happens that, by tacit agreement, firms refrain from 'passing on tc
the consumer* their fall in costs. This is equivalent to an increase in the degree ol
monopoly, and has a tendency to reduce output (see below, p. 94) . For our present
purpose we may assume that the degree of monopoly is unchanged, but the pos-
sibility, highly unfavourable to the scheme, that it may lead to an increase in the
degree of monopoly, must not be left out of account.

8 The increase in consumption will lead to a temporary increase in investment,
which will cause a further increase in employment. Cf. p. 48.

58 ESSAYS IN THE THEORY OF EMPLOYMENT

the base date. His marginal costs will therefore be
lowered. Thus entrepreneurs have an incentive to lower
prices, 1 each one hoping to get custom from his rivals.
But average costs have fallen by very much less, since the
full wages of men formerly employed have still to be paid.
A fall in prices will therefore reduce profits. If the former
output were produced prices would fall, real wages would
rise and less than the former level of savings would obtain.
Consequently output does not remain constant, but
increases.

The expansion of output is greater than would normally
occur in response to the same decline in thriftiness. An
expansion of output, by absorbing workers into employ-
ment, normally reduces the dis-saving represented by dole
payments. But in the present case dole payments cannot
fall below their amount at the base date. Saving will
therefore increase less readily, as output expands, and the
expansion of output will proceed further than in the
normal case.

The increase in consumption due to the scheme will lead
for a time to increased investment, and consequently to a
further increase in employment. When the appropriate
adjustment in the stock of capital has been made, employ-
ment in the investment industries will return to the same
level as would have obtained without the scheme. It is
therefore a legitimate simplification to assume that, after a
period of transition, the volume of investment in real terms
is the same as at the base date. The level of output must
therefore have expanded sufficiently to ensure that, with
reduced thriftiness, the level of saving in real terms is the
same as before. If all wage incomes are spent, and if entre-
preneurs and rentiers are similar, it follows that the level
of real income of entrepreneurs and rentiers, taken together,
must be the same as before.

In this position prices must be lower than before. For if
output expanded so much that, with a lower schedule of

1 If no tacit agreement prevents them; see p. 57, note.

CERTAIN PROPOSED REMEDIES FOR UNEMPLOYMENT 59

marginal costs, prices were unchanged, then there would be
no reduction in net receipts in respect to the old output to
offset net receipts in respect to the increment of output.
Thus at the point where the surplus over wages costs
(allowing for the subsidy) is the same as before, prices
must be lower than before. From this a curious conse-
quence follows. Some firms, which were working close to
capacity at the base date, and which are presumably
the most efficient, will be unable to expand and so will
employ no subsidised labour. These firms will be compelled
by the fall in prices to reduce their output, and may even
be driven to close down altogether. 1 Demand is transferred
from them to firms which are in a position to expand.
This reshuffling of output between firms leads to an
increase in dole payments. For the expanding firms will
now be receiving the subsidy in respect to workers who
were already employed at the base date, but who passed
through a period of unemployment when their original
firms dismissed them. Thus the amount of dole payment
corresponding to a given output will gradually increase
as the process of reshuffling goes on. Thriftiness will
therefore decline and total output increase. In the final
position, a larger total of output will be produced by a less
efficient selection of firms.

The analysis of this scheme may appear somewhat
fantastic, but the fantasy lies in the scheme, not the analysis.

1 Moreover, the fall in prices will increase the burden of fixed indebtedness, and,
since such schemes are never proposed when entrepreneurs are prosperous, the
subsidy may be expected to lead to widespread bankruptcy.

DISGUISED UNEMPLOYMENT

AN economy consisting of self-supporting families each
working their own land must always enjoy full employ-
ment, since each individual is free to work as long as he
considers the real reward he obtains a, sufficient induce-
ment for his efforts.

In any economy in which there is specialisation and
exchange the principle of effective demand comes into
play, and unemployment may occur. If, however, there
are no debts, no form of money and no negotiable capital
instruments, 1 the output of the community will be in
neutral equilibrium. Saving can only be done by means
of adding to the stock of durable goods (an increased
amount of seed corn may be saved from the year's harvest
so as to extend next year's crop) . An increase in thriftiness
is therefore necessarily accompanied by an increase of
investment, and an increased desire to invest necessarily
entails an increase in thrift. A communist economy with-
out private saving is of this type.

Any community with an exchange economy and nego-
tiable capital exhibits the main features of our own system,
and in such a system there is no reason to expect that full
employment will be the normal state. 2 On the contrary,
full employment is only likely to occur in periods of
abnormally rapid expansion, when inventions and dis-
coveries are giving constantly renewed stimulus to invest-
ment. In a relatively stagnant state of society we should
expect under-employment to be the rule. Yet unemploy-
ment, as we know it, is specifically the disease of an
advanced industrial community. How can we account
for the fact that, over the whole range of human history,

1 General Theory, p. 239. a Ibid., p. 347.

60

DISGUISED UNEMPLOYMENT 61

unemplqyment in the modern sense is, comparatively
speaking, a rare and local phenomenon?

The answer is to be found in the existence of disguised
unemployment. In a society in which there is no
regular system of unemployment benefit, and in which
poor relief is either non-existent or 'less eligible 5 than
almost any alternative short of suicide, a man who is
thrown out of work must scratch up a living somehow or
other by means of his own efforts. And under any system
in which complete idleness is not a statutory condition for
drawing the dole, 1 a man who cannot find a regular job
will naturally employ his time as usefully as he may. Thus,
except under peculiar conditions, a decline in effective
demand which reduces the amount of employment offered
in the general run of industries will not lead to 'unemploy-
ment' in the sense of complete idleness, but will rather
drive workers into a number of occupations selling
match-boxes in the Strand, cutting brushwood in the
jungles, digging potatoes on allotments which are still
open to them. A decline in one sort of employment leads
to an increase in another sort, and at first sight it may
appear that, in such a case, a decline in effective demand
does not cause unemployment at all. But the matter
must be more closely examined. In all those occupations
which the dismissed workers take up, their productivity is
less than in the occupations that they have left. For if it
were not so they would have engaged in them already.
The wage received by a man who remains in employ-
ment in a particular industry measures the marginal
physical productivity of a similar man who has been
dismissed from it, 2 and if the latter could find an occupation
yielding hirrv a better return, he would not have waited
for dismissal to take it up. Thus a decline in demand for the
product of the general run of industries leads to a diversion
of labour from occupations in which productivity is higher

1 The *dole' is here used to mean any kind of relief payments.

2 This is upon the assumption of perfect competition in the industry, but the
existence of any degree of monopoly merely adds an a fortiori consideration.

62 ESSAYS IN THE THEORY OF EMPLOYMENT

to others where it is lower. 1 The cause of this diversion, a
decline in effective demand, is exactly the same as the cause
of unemployment in the ordinary sense, and it is natural
to describe the adoption of inferior occupations by dis-
missed workers as disguised unemployment.

In this connection it is convenient to make use of an
alternative definition of unemployment. When full em-
ployment obtains, an increase in the output of investment
goods can only dccur if there is a decline in the output of
consumption goods (any reasonable arbitrary division
being made between the two classes) . On the other hand,
when less than full employment obtains, an increase in
investment will normally lead to an increase in consump-
tion, and a decline in thriftiness will normally lead to an
increase in investment as well as in consumption. Thus we
may say that unemployment is present when an increase
in the output of capital goods (not offset by an increase in
thriftiness) would lead to an increase in the output of
consumption goods. 2

Let us apply this criterion to disguised unemployment.
If a revival of investment were to occur, dismissed workers
would be called back from the hedgerows and the street-
kerbs into their normal occupations. The wages they now
receive represent a command over consumption goods
which they prefer to the product of their former hand-to-
mouth efforts. The output of consumption goods, as
evaluated by consumers, has therefore increased. Hence,
according to our definition, unemployment existed before
the revival of investment took place, even though every
individual worker was busy all day long. There has been

1 We are here confronted with the formal difficulty of distinguishing a divergence
of marginal products of similar workers whjch is due to a decline in demand, from one
which is due to ignorance, inertia, or bad management. But in principle the distinc-
tion is simple, for the first would disappear and the second would not if a revival of
effective demand were to occur.

2 It was suggested earlier (p. 10) that the criterion of full employment is that no
one entrepreneur can increase his staff without reducing the staff of some other entre-
preneur. This criterion gives the same result as the above, provided that the men
employing themselves in inferior occupations are not reckoned as entrepreneurs.

DISGUISED UNEMPLOYMENT 63

no increase in employment reckoned by heads, but there
has been an increase in employment reckoned in terms of
output, because efficient methods of production have been
substituted for inefficient methods. 1

The level of consumption corresponding to a given level
of effective demand will be higher the better are the
opportunities of the unemployed for self-help (assuming
that dole payments are in any case out of the question),
for the product of their efforts, the equivalent of which
they consume themselves, is a clear addition to the output
of the regular industries. The increase in consumption
brought about by re-employing them is therefore less the
smaller is the difference between their productivity in
tegular industry and their productivity in hand-to-mouth
occupations.

In some cases this difference may be slight. When
there is an open frontier, with free access for all comers to
good cultivable land, the difference in productivity
between a man in employment and a man in disguised
unemployment may be small. In the limiting case, where
there is no difference in productivity, unemployment can
never occur, for a man dismissed from industry can then
take up an alternative occupation without any loss in real
earnings, 2 and an increase in investment could only take
place if consumption were to decline.

On the other hand, the less productive are the hand-to-
mouth occupations, the nearer will the unemployed be to

1 The notion of disguised unemployment throws light on an interesting, though
highly academic, problem. It has always been felt that the definition of 'employment'
was arbitrary, and that a man when he is shaving himself, or a woman when she is
scrubbing her own floor, is 'producing utilities' just as much as when he is mining coal
or she is serving at a counter. Wicksteed carried this line of argument to its logical
conclusion, and held that we are producing utilities tor ourselves even when we are
lying in bed. On his view, everyone is 'employed* for twenty-four hours every day.
The analysis of disguised unemployment makes it clear that while everyone is occupied
for twenty-four hours a day, so that the total amount of occupation can never be
increased, yet employment can be said to increase when part of a man's time is trans-
ferred from an occupation in which its productivity is lower to one where it is higher.

8 This approximates to the case of a community of self-supporting families, for this
condition can only be fulfilled when no increase in productivity results from employ-
ment under entrepreneurs in specialised industries.

64 ESSAYS IN THE THEORY OF EMPLOYMENT

starvation, and the stronger will be the pressure upon
society to institute some kind of dole system. The attitude
of mind, prevalent even now in certain quarters, that
unemployment is the result of a vicious idleness of disposi-
tion in the unemployed individuals, pandered to by the
dole, is largely an anachronism which had some plausibility
in an epoch when there was open access to the land, so
that any active and laborious individual could make a
livelihood, when he fell out of employment, not glarijigly
different from what he had obtained in hie former trade.

The existence of disguised unemployment introduces a
complication into the formal scheme of the General
Theory of Employment. When it is possible for unemploy-
ment to become disguised, there is not a unique function
relating total consumption to total investment, since a
given rate of investment will be accompanied by a greater
rate of consumption the more unemployment is disguised.
In the normal way an increase in output is accompanied
(in the short period) by a fall in real wages and an increase
in real profits, in respect to the output which is already
being produced. The increase in profits leads to an
increase in saving, and it is for this reason that, even when
there is no dole, an increase in output, in the normal way,
can come about only if there is an increase in investment.
But when unemployment becomes disguised there is an
increase in output unaccompanied by an increase in saving.
Some workers have found an occupation in which real
earnings are low without there being any increase in profits
and output expands without there being any increase in
saving. 1

1 From a certain point of view the phenomenon of disguised unemployment may
be regarded as a special case of a change in relative wages. In general, when money
wages are falling, for any reason, they are likely to fall unevenly, those workers who are
in a relatively weak bargaining position suffering a relatively large cut in money
wages. There will thus be a fall in the real wages of some workers and a rise in the
real wages of others. Output will expand (or contract by less than it otherwise would
have done) where relative wages fall, and contract by more than it otherwise would
have done where relatives wages rise. A change in relative wages will alter the dis-
tribution of profits between different groups of employers, relative profits rising where

DISGUISED UNEMPLOYMENT 65

2

The notion of disguised unemployment has some rele-
vance even at the present day. Its effects may most
conveniently be examined in two stages. We will consider
what happens when an individual who is out of work takes
up some hand-to-mouth occupation, first when he has no
right to any form of relief, and, second, when having been
drawing the dole (which is taken to stand for all forms of
relief) he is now disallowed benefit.

To simplify th argument we will assume that the dole
is financed entirely by borrowing, so that a reduction in dole
payments is equivalent to a reduction in the central or
local budget deficit that is, to a decline, in dis-saving.
When this condition is not fulfilled, the dole being financed
from rates and taxes, the situation is unaltered for at least

relative wages fall, and falling where relative wages rise. This will have a reaction
upon the thriftiness of the community as a whole which may be in either direction
and may, consequently, lead either to an increase or to a decrease in employment.
Assuming for simplicity that all wage incomes are spent, the condition for a decline
in thriftiness (and an increase in employment) is that saving out of profits declines
by more, in the contracting industries than it increases in the expanding industries
(assuming that the savings of consumers are unaffected by changes in relative prices).
This condition will be fulfilled in so far as (a) the entrepreneurs in the expanding
industries are poorer, and therefore less inclined to save, than the entrepeneurs in
the contracting industries, and (b) the short-period elasticity of supply is greater in
the expanding industries, so that profits in them increase by less than profits in the
contracting industries decline. There is no particular reason to suppose that this
condition will in general be fulfilled, except, perhaps, in so far as workers are likely
to be least strongly organised where employers are poorest; and haphazard changes in
relative wages are as likely to reduce total employment as to increase it.

The condition for the increase in output in the industries where relative wages fall
to be offset by no decline in output in other industries is that there should be no
increase at all in saving out of profits in the expanding industries. This will occur
only if the entrepreneurs engaged in those industries are too poor to save in the first
instance and if the elasticity of supply is so great that their profits do not increase up
to the point at which saving begins. It is clearly unlikely that this more stringent
condition should be fulfilled by any ordinary change in relative money wages. But
when we are discussing disguised unemployment we are supposing that unemployed
individuals are restrained by loyalty, or by the strength of Trade Union organisation,
from competing for jobs in regular industry by offering themselves at cut-wage rates,
and that they take up occupations in which they are able to employ themselves. In
such a case the entrepreneurs in the expanding industries, i.e. the unemployed men
themselves, are extremely likely to be so poor as to spend the whole of their receipts.
It is for this reason that the expansion of their output is not accompanied by any
contraction in the cutout of reerular industrv.

66 ESSAYS IN THE THEORY OF EMPLOYMENT

one financial year; if, after a time, a decline in dole pay-
ments leads to a reduction in taxation, some part of the
increased net income of taxpayers will be devoted to con-
sumption, so that less than the whole of any reduction
in dole payments represents an increase of saving. This
alters the magnitude, but not the direction, of the effect
upon employment of changes in the amount of the dole.
Our assumption that the whole of the dole represents dis-
saving simplifies exposition without introducing any dif-
ference of principle. We will further- assume that an
unemployed man has no saleable wealth, no relations to
help him, and no credit with his tradesmen.

In the first case, the individual we are considering has
no source of expenditure at all and is faced by starvation.
He takes up some occupation planting potatoes if he can
get an allotment, selling match-boxes in the Strand,
hanging round railway stations to carry bags to hotels.
Anything that he earns, in kind or in cash, he devotes to
immediate consumption. Whatever he succeeds in pro-
ducing is a clear addition to the total of output. As a first
approximation we may say that from the point of view of
the rest of society, taken as a whole, his activities make no
difference to output, one way or the other. It may be that
particular producers suffer from his competition. 1 The
tobacconists will complain if he sells matches. But the
money which he attracts from the customers of the tobac-
conists is spent at the grocer's. The total output of match-
selling is not increased by his efforts, but the total output
of society is increased by a value exactly equal to what he
spends. He adds to demand exactly what he adds to
supply, the grocers gain from his addition to demand just
what the tobacconists lose from his addition to supply, and
the output of society, excluding himself, is neither increased
nor diminished. The dole-less individual, who is too poor

1 In so far as he is competing with other men already in disguised unemployment
he is imposing a hardship upon them, and in so far as his output leads to a curtail-
ment of theirs the whole of his output fails to be a net addition to the output of
society.

DISGUISED UNEMPLOYMENT 67

to save, is a little world to which Say's Law applies with
full force.

It is to be observed that it makes no difference to the
argument whether the unemployed man eats up his own
produce or sells it to others. If he sells matches and buys
potatoes, the tobacconist loses what the greengrocer gains,
and if he grows his own potatoes, neither is affected at all.
In either case, the output of the rest of society is, on balance,
unaffected by his activities.

This is a first approximation. There are certain ways
in which his activities may alter the output of the rest of
society, but their effect is likely to be slight. First, it may
be that the ordinary traders with whom he comes into
rivalry are more disposed to saving than those on whom he
spends his earnings. 1 If a decline in the profits of tobac-
conists reduces saving by more than an increase in the
profits of grocers promotes it, then a transfer of custom
(via the street seller) from one to the other will reduce
thriftiness and increase output for the rest of society.
But there is no a priori reason to expect that this effect will
tell in one direction rather than another. 2 Second, charit-
able persons may feel inclined to buy more matches when
they are offered on the kerb than when they are sold only
in the shops, or commodities which are unobtainable in
better times, such as the services of an outside porter, may
attract expenditure from consumers on goods which they
would otherwise forgo. If, but only if, this expenditure
is not offset by economising on other lines of consumption,
a decline in thriftiness is induced by the hand-to-mouth
efforts of the unemployed man. On the other hand, the

1 Or from whom he buys raw materials. The unemployed man may be obliged
to purchase raw materials from the regular industries, but he can only do so from the
proceeds of his sales. He cannot buy seed potatoes and eat the crop himself, since he
has no resources to dis-save. The increase in demand, represented by his purchases
of raw materials and consumption goods together, is offset by the increase in supply
represented by the output which he places on the market.

2 If a large part of the match-seller's earnings are devoted to paying rent, then,
since a landlord is likely to save more than a tobacconist, it is probable that the effect
of the match-seller's efforts will be to reduce the output of the rest of society.

6& ESSAYS IN THE THEORY OF EMPLOYMENT

outside porter may save the traveller from taking a taxi,
and this economy may not lead him to an equivalent
expenditure on something else. 1 It appears that the net
reaction of the efforts of the unemployed man upon the
thriftiness of the rest of the community can hardly be very
great. Our first approximation is a good one, and these
complications may legitimately be neglected in what follows.

We have seen that the self-help of a man who in any
case does not draw the dole makes no appreciable difference
to the rest of the community taken as a, whole, though he
may damage some sections and help others. The case of a
man who is disallowed benefit is not the same. For, as we
have seen, whatever he may now do for a livelihood adds
to demand exactly what it adds to supply; thus whether
he starves, grows potatoes, or sells matches, whether he
now earns more 2 or less than he received as dole, it makes
no difference. Demand for the output of the rest of society
is reduced by the amount of dole that he was formerly
spending. Thus, when he is disallowed from benefit, the
output of the rest of the community will decline. 3 The effect
of discontinuing the dole to an unemployed ma'h, and
driving him to self-help, is to increase the total of unem-
ployment, while causing a part of it to become disguised.

It is now plain that the institution of a dole where none
was before cannot lead to a decrease of employment,
according to our definition. If a dole is instituted, for which
complete idleness is a qualification, the result will be an
increase of employment and output in regular industries,

1 To put the same point in a formal manner: the efforts of the unemployed alter
the composition of the aggregate of goods 'offered to consumers, and so may alter
the eligibility of consuming rather than saving; cf. p. 36.

z This is not inconceivable. If self-help methods provide a better income than the
dole, the unemployed man may forgo benefit voluntarily, and the effect of this is
exactly the same as though he were disallowed. A more plausible case is one in which
a man prefers the dole, so long as he is eligible, to an onerous or undignified line of
life which earns him more money.

8 In the limiting case, where no decline in saving accompanies a decline in output
except the increase in dis-saving due to dole payments, one more man must be thrown
out of wo'rk, and receive the dole, for every unemployed man disallowed benefit;
cf. p. 47, note.

DISGUISED UNEMPLOYMENT 69

combined with a decrease of disguised unemployment.
It will cause unemployment to throw off the disguise, but,
from the point of view of regular industry, the amount of
unemployment will be reduced.

It should be conceded, however, that the introduction
of the dole system must increase the amount of enforced
idleness, of which there is none when all unemployment
is busily disguised, and may, if efficiency in the occupations
which disguise it is sufficiently high, lead to a decrease in
the total of outpfut. Thus, in some circumstances, it is
possible to make out a case against the institution of a
dole for which idleness is a qualification, though not in a
country like our own, where the opportunities for self-help
are exceedingly meagre and efficiency outside ordinary
industry is very low.

But a dole for which idleness is not a qualification is
an unmixed benefit. An example of such a system is
to be found in self-help schemes such as that instituted at
Upholland. Here a community of unemployed men work
at various trades for their own benefit. They continue to
draw the dole to which they were entitled when they became
unemployed, and with this imports into the community
are paid for. Their own produce is not exported (i.e.
sold to the outside world), but is divided up and consumed
within the community. Our analysis enables us to see
that the output of the rest of the world is unaffected by
the existence of such a community, because the dole of an
unemployed man who enters Upholland continues to be
spent upon the output of the outside world, while the
whole produce of the workers within the scheme is a clear
addition to the output of society. Further, we have seen
that the distinction, rigidly drawn at Upholland, between
working for home consumption and working for sale is
somewhat artificial. Entrepreneurs engaged in the par-
ticular lines of production, for instance market gardening,

70 ESSAYS IN THE THEORY OF EMPLOYMENT

developed at Upholland suffer from increased competition,
for without Upholland a larger part of the dole would be
spent upon vegetables. If exports from Upholland were
allowed, this effect would be intensified and no doubt
the outcry that would be raised by the employers (and
perhaps the workers also) in these trades, and the abusive
references to prison labour that they would permit them-
selves, would create so much unpleasantness that it is
scarcely practical politics to advocate exports from Up-
holland. 1 But the fact remains that, so tang as all receipts
from exports were currently spent upon imports from the
outside world into Upholland, the outside world as a whole
would be no worse off. Certain trades would suffer from
increased competition, and certain other trades would gain
from increased demand. On balance, Upholland would
gain from increased variety of consumption, and the rest
of the world would be no worse off.

The Upholland scheme is not in itself a remedy for
unemployment. It is no substitute for measures calculated
to increase effective demand. But it provides a peculiarly
efficient method of reducing the ill-effects of unemploy-
ment, for it allows the advantages of hand-to-mouth
production, under unusually favourable conditions, to be
obtained without imposing, upon the unemployed and
the rest of society alike, the evils of cutting off the dole.

Under ideal conditions, the widespread institution of
such schemes could remove the ill-effects of unemployment
altogether, and could produce the level of consumption
corresponding to full employment, or even improve upon
it. Suppose that the entrepreneurship, provided by well-
wishers of the unemployed, does not compare unfavourably
in efficiency with the ordinary run of entrepreneurship
produced by the pursuit of profit, and suppose that the

1 In such cases, the interests of those who suffer by competition are more concen-
trated, and therefore more vocal, than the interests of those who gain by increased
expenditure. If the firms who would gain by exporting to Upholland could be taught
to cry as loudly as the firms who lose by the competition of exports from it, the neces-
sity of an embargo on exports from Upholland would disappear.

DISGUISED UNEMPLOYMENT 71

schemes work on a large enough scale to approximate
to the most efficient technical methods of industry. Then
the productivity of a man in an Upholland community
would not be much less than that of a man in ordinary
industry. The level of consumption corresponding to full
employment would be attained, provided that everyone
thrown out of ordinary industry entered an Upholland
community, if efficiency inside bore such a relation to
efficiency outside that earnings within the community
exactly made up the difference between the dole and an
ordinary wage. A man who lost his job would then enjoy
the same standard of life as a man who retained it, and an
increase in investment would, at best, lead to no increase
in the total of consumption. 1 If productivity in Upholland
stood in a higher ratio to productivity in ordinary industry
than the ratio of wage minus dole to wage, then the total of
consumption would necessarily increase as ordinary em-
ployment declined, and the workers would actually gain
by a decline in effective demand. This apparent paradox
is merely due to the fact that workers are imagined to be
drawing the dole on top of earnings not much less than
those obtained in ordinary industry.

All this is far away from the conditions of Upholland
as it actually exists. But our cloud-castle supplies an
important warning which must not be neglected in the

1 This is true in the limiting case, where no change in saving or dis-saving accom-
panies a change in the output of industry except the change in dole payments. Take,
by way of illustration, the dole equal to one-third of the wage. Now suppose that a
man is called out of Upholland and set to work in an investment industry. His total
consumption is unaltered, but whereas formerly only one-third of it was drawn from
the consumption industries outside Upholland, now the whole of it is drawn from them.
The multiplier for the outside world is equal to 3, and two men are put into work in
the consumption industries, an increase in investment equal to one man's wage lead-
ing to a decrease in dis-saving of three men's dole. Now three men have been with-
drawn from the production of consumption goods in Upholland, and, of these, two
are producing consumption goods outside. Thus if their efficiency in Upholland is
two-thirds of their efficiency outside the total of consumption is unchanged. Invest-
ment could then continue to increase, without a decline in consumption, up to the
point at which all workers are reabsorbed from Upholland into ordinary industry.
Full employment is then reached, and any further increase in investment would only
be possible if consumption declined.

72 ESSAYS IN THE THEORY OF EMPLOYMENT

actual situation. As the scope of a scheme such as that at
Upholland expands and its managers gain in experience,
the efficiency with which its output is produced will
increase. A day may not be far distant when, taking
account both of the standard of consumption and the
general amenities, the life of an unemployed man in Up-
holland may appear preferable to the life of an employed
man outside. If things ever came to this pass, it is easy
to imagine the protests that would be made against
continuing dole payments to men who were actually find-
ing unemployment tolerable. But if the dole were dis-
continued, the whole merit, from the point of view of out-
side industry, of the Upholland scheme would disappear.
The increase in efficiency which we have presumed to
occur might be great enough to compensate the men
within the scheme for the loss of dole payments which
we have imagined to be the consequence. But there is
nothing to compensate the outside industries for the loss
of their export trade to Upholland. Unemployment out-
side would increase, 1 and profits decline. The case of
Upholland without the dole is parallel with the' case,
examined above, of the man who was disallowed benefit
and took to selling matches. The men in Upholland are
less wretched than the matchseller. But this is no comfort
to employers in the outside industries who suffer from the
shrinkage of demand.

4

We may now consider, in the light of Upholland, the
effects of the regulations surrounding eligibility for unem-
ployment benefit. We have seen that, provided he does not
sacrifice his right to the dole, an unemployed man does
himself good, and on balance does others no harm, by

1 In the limiting case one additional man would become unemployed for every
one who lost benefit by entering Upholland (see p. 68, note) . If this condition were
fulfilled at all levels of output, the result would be that ordinary industry would
come to a standstill while the whole working population was employed in Upholland
communities.

DISGUISED UNEMPLOYMENT 73

occupying himself as usefully as he can. Yet the regulations
surrounding the receipt of the dole militate strongly against
useful occupation. Until recently an unemployed man
was compelled to fill his time in keeping up the appearance
of Genuinely Seeking Work. If he fulfilled this obligation
conscientiously he had little time or energy for any other
activity. Even now he cannot embark upon any occupa-
tion which commits him to spending a certain period of
time in any definite way, for he is obliged to hold himself
ready to accept *a job, should one chance to offer, at a
moment's notice. But there is another consideration which
is by far the most important. The unemployed man is
hedged in by restrictions on his daily activity which are
due to fear of losing his right to benefit. The restrictions,
in the nature of the case, are highly arbitrary. Under the
regulations at present [1936'] in force 1 a man may pursue a
subsidiary occupation, provided that (a) it is not his usual
trade, (b] it does not occupy what would be his working
hours if he had work, and (c) it does not bring in an income
of more than three shillings and fourpcnce a day. Thus an
unemployed waiter is allowed to accept an odd job in the
morning, and an unemployed shop assistant may work
after six-thirty (provided that neither accepts three and
fivepence for his services), but during normal working
hours they must sit at home doing their best to appear
'capable of and available for' the work that they cannot
find in their normal trades. 2 Such regulations restrict the
disguising of unemployment within narrow limits. More-
over, the extreme complexity of the rules and the Gilbertian
situations to which they give rise have a strong effect in
discouraging any efforts at self-help, for the unemployed
man, bewildered by the intricacy of the regulations with
which he is faced, and discouraged by the prospect of
'coming up for trial' before the Court of Referees, is

1 Unemployment Insurance Act, 1935, Section 35, (5).

2 The concession is made that the subsidiary occupation may actually be carried
out during normal working hours provided that it could be performed outside them
(Emmerson and Lascelles, Guide to the Unemployment Insurance Acts, p. 55).

74 ESSAYS IN THE THEORY OF EMPLOYMENT

naturally inclined to pursue a policy of Safety First and to
sit at home making sure of his right to benefit. 1

Yet, as we have seen, the situation of the rest of the
community is in no way impaired if an unemployed man
succeeds in earning some money, whether less or more
than three and fourpence. Anything, in kind or in cash, 2
that he may be able to secure is a pure gain to himself
and no loss to others, provided that he does not sacrifice
his right to the dole. It follows that regulations calculated
to prevent him from doing himself any <kind of good are
harmful to him and not beneficial to the rest of society,
while regulations that deprive him of the dole on a slight
pretext are deleterious to the rest of society as well as to
himself. The administrative complications involved, the
strong moral objection to scroungers, felt by workers as
well as by taxpayers, and the difficulty of preventing em-
ployers from obtaining an illicit subsidy, may be regarded
as sufficient justification for such regulations. But their
economic effects can only be harmful

1 These regulations apply to eligibility for unemployment insurance. The regula-
tions surrounding relief from the Public Assistance Committees appear to be still
more arbitrary. Sir John Jarvis reports (The Times > June 16, 1936) that some un-
employed men in Jarrow, working on a purely co-operative basis to make a sports
ground for themselves, were informed that they would be disqualified for public
assistance relief if the work continued, although men entitled to unemployment in-
surance were permitted to take part in the scheme. There was no element of profit
in the scheme, and the workers engaged on it received no pay except a mid-day meal,
and boots and trousers in which to work, provided by the Commissioner for Special
Areas.

2 It must be recalled that we are assuming throughout that an unemployed man
spends all he earns. In practice some part of his cash earnings may be devoted to pay-
ing off debts. In that case the increase in supply, represented by what he sells, is not
offset by an equal increase in demand. The grocers, to return to our first example,
do not gain as much as the tobacconists lose. This consideration introduces a difference
between the case where he earns cash and where he works for his own consumption.
But it would be hard to defend the regulations at present in force on the ground that
they compel the unemployed to incur the maximum amount of debts.

PART II
THE LONG-PERIOD THEORY OF EMPLOYMENT

MR. KEYNES' General Theory of Employment has been
developed mainly in terms of short-period analysis, and
the background of equilibrium theory which corresponds
to it is largely unexplored. The purpose of this essay is to
outline a method by which Mr. Keynes' system of analysis
may be extended into the regions of the long period and
by which it may become possible to examine the long-
period influences which are at work at any moment of
time. Only a small section^ of the subject is here discussed,
and the discussion is carried on in the highly abstract
terms which are inevitable in the early stages of such an
inquiry. But the importance of the subject is not merely
academic and even a highly formalised treatment of it may
be wprth the attempt.

Consider a closed community, living under a capitalist
system, with population stable in respect to numbers and
to age distribution, and with given tastes and technical
knowledge. The problem with which we shall be mainly
concerned is the effect upon equilibrium conditions in
such a community of changes in the rate of interest.
Let us first suppose that a certain rate of interest has been
established and is maintained at an unvarying level. 1
In this situation, provided that the given conditions have
endured for sufficient time, net investment will have ceased.

1 The static conditions here postulated most not, of course, be taken to present
a picture of a situation which would ever be likely to occur in actuality. They are
adopted as a theoretical construction which may enable us to isolate certain tendencies
whose influence in any actual situation are disturbed by innumerable cross-currents.
The conception of equilibrium employed in this essay is the Marshallian conception
of a position of rest towards which the system is tending at any moment.

75

76 ESSAYS IN THE THEORY OF EMPLOYMENT

For, as long as capital goods continue to accumulate, their
profitability at the margin declines and the incentive to
further investment is continuously weakened. Investment
is always tending to bring itself to an end, 1 and in the
stable conditions that we are considering nothing happens
to revive the inducement to invest as it flags. In conditions
df equilibrium the stock of capital is adjusted to the given
rate of interest, and no further accumulation takes place.
The marginal efficiency of capital corresponding to zero
net investment is equal to the rate of interest and if, by
chance, positive or negative investment were to occur,
the marginal efficiency of capital would cease to be equal
to the given rate of interest. If new investment were to
take place capital would be increased and its earnings at
the margin would fall. The marginal efficiency of capital
would then be less than the rate of interest. The investment
would turn out to have been unprofitable, capital goods
would not be worth replacement and a movement back to
equilibrium would set in with a decline in the stock of
capital. On the other hand, if, in equilibrium, the stock of
capital goods were allowed to deteriorate the marginal
efficiency of capital would rise above the rate of interest
and investment would take place until the stock of capital
was restored to its former level. 2 The familiar phrase
'long-period equilibrium' may be adopted to describe this
situation. For, in the ordinary sense, a firm or an industry

1 Assuming that the proportion of total income saved rises with income. This is
the natural assumption to make (see Harrod, The Trade Cycle, pp. 106-9). If saving
were proportional to income over the whole range of income, any initial impulse in
the upward direction would lead to continuous expansion (Harrod, o/>, cit. 9 p. 90),
while zero net investment would entail zero total income. The propensity to save
which is here in question must not be confused with the propensity to save under
conditions of long-period equilibrium discussed below, p. 77 and pp. 10^2-111.

2 In the static conditions that we are considering, equipment is conceived to be
wearing out at a steady rate, each capital instrument, in equilibrium, is renewed as it
wears out, production proceeds in a continuous flow, and to-day is merely a repetition
of yesterday. In such conditions the logical difficulty inherent in the notion of a con-
stant stock of capital does not arise. A change in the ratio of capital to labour is likely
to be accompanied by a change in the nature of the capital goods employed, but so
long as tastes and knowledge are unaltered there is no ambiguity about the direction
of a change in the amount of capital.

THE LONG-PERIOD THEORY OF EMPLOYMENT 77

which is under an inducement to expand or contract the
plant which it employs is not in equilibrium, and long-
period conditions are established only when investment
has come to an end. 1

The short-period analysis of the level of employment
can readily be adapted to describe the forces which deter-
mine employment in our static community. The rate of
interest determines the amount of capital per unit of
labour employed. 2 To each level of employment, therefore,
there corresponds a certain long-period level of total out-
put. The rate of saving corresponding to a given total
output depends, if the propensities of individuals to save
are given, upon the distribution of total income between
classes. Let us postulate that the degree of monopoly in
the economic system is giv^n, that the institutional factors
governing the distribution of wealth are stereotyped and
that fiscal policy is not altered. Then to each level of
total output will correspond a given system of distribution
of real income between classes. 3 Thus to each level of out-
put there corresponds a certain rate of saving, which must
be imagined to occur if that level of output is imagined to
obtain. It is possible to draw up a schedule relating each
level of output with the amount of saving (in real terms)
which would take place if that output were attained under
long-period conditions. This schedule depicts the thrifti-
ness of the community at the given rate of interest in long-
period conditions. It is natural to assume that a higher

1 Mr. Keynes (General Theory, p. 48, note) uses 'long-period equilibrium' in a
slightly more extended sense. My long-period equilibrium is a special case of Mr.
Keynes' equilibrium, and in my terminology Mr. Keynes' position of long-period
equilibrium with investment going on is a situation in which the equilibrium position
is moving ahead of the actual position at a steady and foreseen rate.

2 In order to simplify the argument it may be assumed that the hours, intensity
and personal efficiency of work are stereotyped and that relative wages are constant,
so that a man-day of standard labour may be taken as the unit of employment (cf.
General Theory 9 p. 42).

8 For long-period analysis it is natural to abstract from the changes in distribu-
tion which are brought about by changes in prices when certain incomes are fixed
in money by long-term contracts.

78 ESSAYS IN THE THEORY OF EMPLOYMENT

level of saving will correspond in the schedule to a higher
level of output. 1

Now, in equilibrium the rate of net investment is
zero. There is therefore only one level of total output
which will give equilibrium the output at which net
saving is zero. Any addition to total output above this
level would bring with it, in long-period conditions, a
smaller increase in total consumption, for part of the
increment of incomes would be devoted to saving. Supply
would have increased more than demand, and equilibrium
would not obtain. Conversely, if total output were to
fall below this level demand would have fallen less than
supply and equilibrium would be restored by an increase
of total income. 2 With zero investment, output, consump-
tion and income, for the community as a whole, are
synonymous.

We have now seen how employment is determined in
the long period. There is one level of output at which
saving is zero, and one level of employment corresponding
to that level of output. The level of employment, deter-
mined in this way, is by no means the same thing 'as the
amount of work which the community is willing to per-
form, and there is no reason to suppose that our static
community will be free from unemployment. If, at the
level of total real income corresponding to full employ-
ment, the rate of saving would be greater than zero, that
level of total income cannot be attained. 3 There is there-
fore no guarantee that full employment will be secured.
It is likely, however, that the larger the amount of un-
employment corresponding to a given level of employment

1 See below, p. 102, for a discussion of the conditions necessary to validate this
assumption.
, a Cf. General Theory, p. 30.

8 On the other hand, in a poor or spendthrift community, the level of real income
corresponding to full employment, at a given rate of interest, may produce a negative
rate of saving. In such a case the equilibrium level of employment would exceed the
available supply of labour, which is an impossible situation. The solution of the prob-
lem lies in the fact that the rate of interest cannot be maintained at a value at which
the demand for labour overruns the limits set by the available supply, for the attempt
to maintain it would lead to cumulative inflation; cf. p. 17. ,

THE LONG-PERIOD THEORY OF EMPLOYMENT 79

the lower will be the schedule of thriftiness. For the un-
employed must somehow be provided for, and their
consumption is likely to be made to some extent at the
expense of the savings of the rest of the community. 1 Thus,
other things equal, the level of employment is likely to be
higher the larger is the available supply of labour. But
the amount of unemployment, in equilibrium, must be
whatever is sufficient to reduce net saving to zero for the
community as a whole.

It is to be observed that, in equilibrium conditions, a
spontaneous increase in the desire to save will reduce the
level of total output, for it will reduce the level of output
corresponding to zero saving. Moreover, it will reduce
the total stock of capital that will be maintained at a
given rate of interest. For with a reduction in output
there will be a reduction iri employment, and, if the stock
of capital were maintained, there would be a rise in the
marginal productivity of labour and a fall in the marginal
efficiency of capital. But if the rate of interest is unchanged
the marginal efficiency of capital cannot alter. Therefore,
in face* of an increase in thriftiness the stock of capital will
decline to the point at which, with zero net investment,
its marginal efficiency is restored to the former level.
Thus thriftiness tends to reduce the size of the equilibrium
stock of capital. 'This was sometime a paradox, but now
the time gives it proof. 3

We may now consider the change in the position of long-
period equilibrium corresponding to an alteration in the
rate of interest.

In the short period a fall in the rate of interest (provided
its influence is not offset by some other change) will
necessarily lead to an increase of employment and of total
income. 2 But this increase of income can only last as long

1 See p. 47 and p. 109.

2 Unless full employment already obtains, in which case the reduction in the rate
of interest cannot persist.

8o ESSAYS IN THE THEORY OF EMPLOYMENT

as the investment which causes it, and meanwhile other
influences come into play. There are three aspects of the
accumulation of capital. While investment is going on
it causes an increase in effective demand, and from a
strictly short-period point of view this is the only aspect
of accumulation that is important. The increase in
current total income brought about by building a Tower
of Babel is just as great as the increase brought about by
investing an equal sum in electrifying a railway system.
As soon as we overstep the narrowest -'boundary of the
short period a second aspect of accumulation must be
brought into account. The greater the extent to which the
amount of capital has increased in any period, the lower
will be the marginal efficiency of capital, with a given rate
of investment, in the next period, and the harder will it
be to maintain a given rate of investment. 1 But this also
belongs to the field of short-period analysis. The third
aspect of accumulation is the effect of an increase in the
stock of capital upon the rate of consumption, and so of
employment, corresponding to a given rate of investment;
it is through this channel that the specifically long-period
effects of accumulation begin to be felt as soon as invest-
ment has proceeded for a certain time. In a discussion of
equilibrium conditions the influence of the current rate of
investment upon effective demand disappears from the
picture, since in equilibrium investment is equal to zero,
and it is with the long-period effects of a fall in the rate of
interest upon consumption that we are alone concerned.
At first sight there appears to be a strong contrast
between the part which the rate of interest plays in the
short-period Theory of Employment and the part which
it plays in the traditional long-period Theory of Distribu-
tion. For instance, in the short period a fall in the rate of
interest will lead to an increase of output, and since with
fixed equipment the marginal productivity of labour
declines as employment increases, a fall in the rate of

1 General Theory > p. 106.

THE LONG-PERIOD THEORY OF EMPLOYMENT 81

interest is associated with a fall in real wages. But the
increase in capital per head consequent upon a lower
rate of interest is likely, in the long run, to raise the level of
real wages. Or, a fall in the rate of interest is associated
with an increase in employment, with given plant, so that
the proportion of labour to capital rises when the rate of
interest falls, but in the long run a fall in the rate of interest
tends to reduce the proportion of labour to capital. It is
one function of the long-period Theory of Employment to
reconcile this apparent contradiction, and to fit the propo-
sitions of the traditional Theory of Distribution into their
place in the analysis of employment.

The problem with which we are faced is somewhat
intricate, and in order to reduce it to manageable propor-
tions it is necessary at the first stage to introduce a clastic
simplification. We will exa'mine first a case in which land
and entrepreneurship are superabundant, so that in effect
there are only two factors to consider labour and capital,
and we will suppose that constant physical returns prevail,
in the sense that an equal proportional increase in labour
and capital will produce the same proportional increase
in output. Further, it is convenient to make the traditional
assumption that output is divided into a number of distinct
commodities, the production of each of which is conducted
under conditions of perfect competition.

Let us suppose that a fall in the rate of interest takes
place and remains in force for sufficient time to allow
equilibrium, with zero investment, to be re-established.
What will be the effect upon the equilibrium level of output?

The first point to be considered is the reaction of lower-
ing the rate of interest upon the desire of individuals to
save. The effect will differ from one individual to another,
according to their circumstances and their psychology.
Some will save more, out of a given real income, when the
rate of interest falls, others will save less. 1 If, for the

1 In other words, a curve connecting the rate of interest with the rate of saving
from a given individual income may be either rising or backward rising.

8a ESSAYS IN THE THEORY OF EMPLOYMENT

community as a whole, a fall in the rate of interest reduces
the desire to save, the total real income at which net saving
is zero will tend to be greater the lower the rate of interest.
But it may be, at least until very low levels are reached,
that every fall in the rate of interest will lead to an increased
desire to save. In this case a fall in the rate of interest will
tend to reduce total income.

To consider every alternative at each stage in the argu-
ment would be wearisome and we will assume for the
moment that the direct effect of a change in the rate of
interest upon the desire to save is, on balance, neutral,
so that the amount saved by the typical individual out of a
given real income remains the same whatever the rate of
interest. Then, if the rate of interest had no other influence
upon t thriftiness, there would be only one level of total
real income at which saving is zero, and a fall in the rate of
interest would leave total real income unaltered.

But there is a further effect to be considered. A fall in
the rate of interest will increase capital per head, so that a
given output is produced by fewer workers using more
'roundabout' methods. The marginal physical produc-
tivity of capital will be reduced and of labour increased,
and the rate of real wages will rise. Now a change in the
distribution of income between workers and capitalists
will have an important effect upon the thriftiness of the
community as a whole. It may be postulated that in our
community the capitalists are a small, and the workers
a numerous class, while the shares of labour and of capital
in the total income are not widely different. The capitalists,
in short, are much richer individuals than the workers, 1
and are consequently more addicted to saving. It follows
that any change in distribution which increases the share
of labour in a given total income will reduce the amount of
saving corresponding to that level of income.

1 It is not, of course, necessary to suppose that every capitalist is richer than every
worker. All we require is that the typical capitalist income should be considerably
greater than the typical worker's income.

THE LONG-PERIOD THEORY OF EMPLOYMENT 83

If a fall in the rate of interest produced this effect, the
equilibrium level of total income would increase as the rate
of interest fell, for the income corresponding to zero saving
would be greater the lower the rate of interest. But it is by
no means necessary that a fall in the rate of interest should
have a favourable effect upon the share of labour in total
income. The rate of earnings of labour is increased, but
the amount of labour employed per unit of output is
reduced. The r^te of earnings of capital is reduced, but
the amount of capital per unit of output is increased.
Thus two contrary tendencies are at work, and the net
result may be either an increase or a decrease in the
income of labour corresponding to a given total income.

Upon the simple assumptions which we have made the
result can be formulated in terms of the elasticity of
substitution. If the elasticity of substitution between labour
and capital is less than unity, the proportional reduction
in labour per unit of output corresponding to a small fall in
the rate of interest will be less than the proportional increase
in the 'rate of real wages, and the share of labour in a
given output will be increased. 1 In this case a fall in the
rate of interest will reduce the amount of saving correspond-
ing to a given total income, the level of income correspond-
ing to zero saving is raised, and the equilibrium income
consequently increased. But if the elasticity of substitution
is greater than unity, labour will lose on the roundabouts
more than it gains on the swings, the amount of saving
from a given total income will increase with the increased
share of the capitalists, the level of income corresponding
to zero saving is reduced, and the equilibrium income
consequently declines.

We have now reached the conclusion that the equili-
brium level of total output will tend to be raised or
lowered by a fall in the rate of interest, according as the

1 See Hicks, Theory of Wages, p, 117.

84 ESSAYS IN THE THEORY OF EMPLOYMENT

direct effect of the fall in interest upon the desire of indi-
viduals to save is negative or positive, and according as
the elasticity of substitution between labour and capital
is less or greater than unity. The effect upon the stock of
capital and employment has still to be considered.

A fall in the rate of interest will necessarily increase the
equilibrium stock of capital, provided that the direct effect
upon the desire to save is not highly positive. In those
cases in which output is increased, the stock of capital is
increased a fortiori. But even if output is reduced it is
reduced precisely because the capitalists have been en-
riched, and since the rate of interest has fallen the level of
income of capitalists can only be raised if the stock of capital
has increased. If, however, a fall in the rate of interest
leads to a sufficiently great increase in the desire of indi-
viduals to save it may actually be associated with a decline
in the stock of capital.

In those conditions in which total income is reduced by a
fall in the rate of interest, employment will be reduced a
fortiori, for not only is there less output, but there is less
employment per unit of output. But in those cases in which
total output is increased it does not necessarily follow that
employment is increased. The diminution in employment
per unit of output due to the increase in capital per head
may more than offset the increase in output, so that the
larger total output may be produced by fewer workers.

It may be convenient to provide a formula to represent
the contrary pulls of increased total output and increased
output per head upon the amount of employment. We
have already discussed the manner in which output alters
in response to a fchange in the rate of interest. Imagine
that a curve is drawn up connecting the rate of interest
with the equilibrium level of total output. Let the elasticity
of this curve be 0. This elasticity, as we have seen, involves
a complexity of factors and must be regarded as a useful
shorthand term rather than as a concept which is of interest

THE LONG-PERIOD THEORY OF EMPLOYMENT 85

in itself. Let c be the rate of interest and O the total
output.

Then = - -r-
dc

c

The proportional increase in employment, due to a small
change in the rate of interest, which would come about if

the ratio of the factors was unchanged, is equal to Q. .

c

The proportional increase in employment per unit of out-
put is equal to - -^ where n is the elasticity of substitution
k'c

and k is the ratio of the cost of labour to the cost of capital. 1
Thus the proportional increase in employment due to a

change in the rate of interest is equal to - !(## - *f).

/* *
(/ i\i

In the simplest case, where the direct effect of changes in
the rate of interest upon the desire to save is neutral and
the elasticity of substitution is equal to unity, there is, as
we have seen, no increase in total output in response to a

1 Let G be the amount of capital, and L the amount of labour, per unit of out-
put, and / the rate of wages.

Now CAc-|-LAJ=o,

and cAC-KAL=o.

*-

AC<* AL AL / /L \

~r~ r "Tl I +7<7

, L* Li JL \ c\-tJ

Ac A/~~ Ac

7 ~ 7 "c

AL

~ Ac cG

c
AL

Jb.0,5
Ac k

c

86 ESSAYS IN THE THEORY OF EMPLOYMENT

fall in the rate of interest. Thus 6 is equal to zero, and
there will be a proportional decline in employment equal
to the proportional fall in the rate of interest divided by
the ratio of labour to capital.

If it is true that 'the most plausible estimate of the
elasticity of substitution is unity,' 1 it appears by no means
fantastic to suppose that a fall in the rate of interest may
reduce the equilibrium level of employment. It is to be
observed, however, that a given fall in the rate of interest
will require a larger increase in the stobk of capital the
greater is the elasticity of substitution. The beneficial
short-period effect of a fall in the rate of interest upon
employment thus tends to be greatest when the long-period
effect is most deleterious.

3

We must now meet the argument that it is unnatural
to suppose that the rate of interest can be permanently
maintained at a level at which unemployment occurs.
It is sometimes held that so long as unemployment exists
forces are set at work tending to depress the rate of interest.
First, it may be argued that unemployment will naturally
lead to a fall in money wages. The direct effect of a fall
in money wages is merely to produce a corresponding fall
in prices and does not by itself have any influence upon
the level of employment. 2 But its indirect effects through

1 Champernowne, Economic Journal ,June 1935, p. 255. Professor Douglas comes to
what is in effect the conclusion that the elasticity of substitution between labour and
capital in the U.S.A. is equal to unity (Theory of Wages, p. 133 etseq.) though he does
not specifically use the conception of elasticity of substitution and, indeed, seems to
have misunderstood its nature (op. cit., p. 59). Evidence for unit elasticity is also pro-
vided by the apparent constancy (over long periods) of the share of labour in total
income (see Douglas, op. cit., p 221, Hicks, loc. cit., p. 131). But, of course, there is
a wide gap between conclusions drawn from our static community and conclusions
applicable to nations of the real world.

8 General Theory, p. 262. An alteration in relative wages may, of course, lead to
some change in employment and an expectation that future wages will differ from
present wages will have important consequences (loc. cit., p. 263). The above argu-
ment applies in its simplest form only to an equal all-round fall in wages which is
expected to be permanent.

THE LONG-PERIOD THEORY OF EMPLOYMENT 87

reactions upon the monetary factors in the situation
cannot be neglected. A fall in prices leads to a reduction
in the demand for money. If the monetary authorities
wish to keep the rate of interest unchanged they can do so
by reducing the supply of money correspondingly, but
if the supply of money is held constant the rate of interest
must fall. 1

There is thus some basis for the argument that the
assumption of' a constant rate of interest is incompatible
with the existence of unemployment, that the position which
we have described is not one of final equilibrium, and that
the level of money wages, and with it the rate of interest,
will continue to fall either until unemployment disappears,
or until either the rate of interest, or the level of money
wages, reaches a point below which it can fall no further.
When one of these three points is reached the tendency for a
fall in money wages to drag down the rate of interest must
come to an end. But we have discovered that it is not
necessarily always true that the first point must be reached

before the second or third, for we have found that in some

* .

cases a fall in the rate of interest merely increases the long-
period level of unemployment. In a community with
perfectly plastic money wages the level of prices may be
always moving toward zero without setting up any tendency
permanently to reverse the situation which is causing
prices to fall. It is thus impossible to argue that there is
any self-righting mechanism in the economic system which
makes the existence of unemployment impossible, even in
the longest of runs.

At best the process of forcing down the rate of interest,
even with highly plastic wages, would be both slow and
uncertain in its operation. A fall in the rate of interest
would be followed by a period of investment, and while
investment continued the level of employment would be
raised. The pressure on money wages would consequently
be relaxed. And if the rate of investment were ever great

1 General Theory, p. 171.

88 ESSAYS IN THE THEORY OF EMPLOYMENT

enough to carry the community temporarily to the point of
full employment, a rise in money wages would be likely
to occur. 1 Thus the run required to reduce the rate of
interest to a given extent, by this route, is likely to be far
longer than the period in which equilibrium to a given rate
of interest can be established. In short, on the assumption
of perfectly plastic wages, our position of long-period
equilibrium with a given rate of interest exists within a
longer-still period, in which the rate of iriterest is deter-
mined by the level of money wages, ancl we have found
that even in the longer-still period unemployment may not
be tending to disappear.

The assumption of perfectly plastic money wages is
highly unrealistic. A community in which money wages
fall without limit so long as unemployment exists is very
unlike the real world, even the pre-trade-union world,
and the absurdity of contemplating a system in which
prices are always moving towards zero is merely the result
of the unnatural assumption on which it is based. 3

But an alternative line of argument presents itsplf. If
the monetary authorities in our static community are in a
position to control the rate of interest by acting upon the
quantity of money, 4 it may be held that in the face of

1 Sec p. 9. There is no contradiction in contemplating a world subject to a chronic
tendency to unemployment in which at the same time there is a secular rise of wages
and prices. The underlying tendency to force wages downwards may be more than
counterbalanced by the upward pressure of a short burst of high investment. An
earthquake, a war, or a major invention may undo in a year the work of centuries
of chronic unemployment. This consideration reinforces the argument advanced by
Mr. Keynes in the General Theory > p. 307.

2 'If wages are once raised, they will never get down again.* Dr. Johnson. (Boswell:
Journal of a Tour to the Hebrides) Tuesday, s8th September).

Cf. p. 6.

4 In a primitive system with metallic currency the quantity of money is fixed by
the stock of the precious metal, and a fall in the rate of interest can only be imagined
to occur through a fall in money wages. When money wages fall the price of the metal,
alone of all prices, remains unchanged and mining is encouraged. Thus the tendency
for >a fall in money wages to lower the rate of interest is reinforced the supply of
money increasing as the demand for it falls. But there is some awkwardness in fitting
the occupation of mining into completely stationary conditions. If yesterday is like
to-day in every respect except that to-day has inherited a larger accumulation of gold,
to-day's rate of interest will be lower than yesterday's. But there is little profit in
studying this unnatural cross between a stationary and an actual world.

THE LONG-PERIOD THEORY OF EMPLOYMENT 89

unemployment they will be likely to adopt a policy of easy
money. Actual monetary authorities are hampered by the
necessity to protect the international equilibrium of their
system. The monetary authorities of a closed community,
though free from this difficulty, may be restrained from
lowering the rate of interest by a tender regard for banking
profits or for the interests of the rentier class. But we will
suppose that their policy is directed towards reducing un-
employment so* far as it is possible for them to do so. In
any given situation a temporary increase in employment
occurs when the rate of interest is lowered and investment
set on foot. Thus the monetary authorities in our static
community will be under recurring pressure to reduce the
rate of interest whenever equilibrium with a given rate
entails unemployment.

But except during the passing phase of positive net
investment an increase in employment is not necessarily a
result of a fall in the rate of interest, for, as we have seen,
there are three stages at which there may be a slip between
the cup and the lip.

First, a fall in the rate of interest may increase the desire
to save, and so tend to reduce total income. Second, the
change in distribution may be unfavourable to labour,
and so tend to reduce total income. Third, even if total
income increases, employment may be reduced, because
of the increase in output per head. In certain communities,
therefore, a policy of lowering the rate of interest in order
to reduce unemployment will frustrate itself. The tem-
porary relief obtained while investment is at its height will
gradually dwindle and the last state will be worse than the
first. Each burst of investment, as the rate of interest is
gradually reduced, will leave behind it the legacy of an
enhanced mal-distribution of income and an increased
level of output per head. A high temporary level of em-
ployment becomes progressively harder and harder to
attain, and the equilibrium level of employment sinks
further and further.

go ESSAYS IN THE THEORY OF EMPLOYMENT

In such a case, " a rise in the rate of interest may be
advocated. But to attempt to cure unemployment by
raising the rate of interest would present itself at any
moment as a very paradoxical policy. For the immediate
effect of a rise in the rate of interest would always be to
cause disinvestment and to increase unemployment. It is
only after the lapse of time that a decline in the stock of
capital could make its influence felt in reducing thriftiness,
by impoverishing capitalists, and in raising" the amount of
labour required for a given output. The most devoted
apostle of long-run benefits would find it hard to advocate
the increased distress which would have to be endured
before any advantage began to appear. Moreover, the
short-period situation is always easier to diagnose than the
long-period, and even the well-known hardihood of
economists trained in the school of equilibrium analysis
might not be sufficient to make them reject a bird in the
short-period hand for a pair of which they may have
managed to catch a glimpse in the long-period bush. At
best, the long-period benefit of a policy of raising the rate
of interest is dubious. The rate of real wages for the em-
ployed workers would be reduced by it. And apart from
the possibility of a negative reaction on the desire to save,
a rise in the rate of interest can increase the equilibrium
level of employment only at the expense of other long-run
advantages. In so far as it produces its effect by reducing
the amount of capital per head, it is keeping productivity
at a lower level than might be attained and curing unem-
ployment merely by 'making work'. On the other hand,
in so far as it increases employment by reducing the share
of capitalists in total income it provides very superficial
treatment for a deep-seated disease. The most effective
remedy for a community which finds itself in such a
situation is to make a direct attack upon the mal-distribu-
tion of income which is the cause of excessive thriftiness.

But even if the policy of raising the rate of interest were
adopted, it could not be guaranteed to secure full employ-

THE LONG-PERIOD THEORY OF EMPLOYMENT 91

ment, for it is likely that there will be a limited range of
values over which a rise in the rate of interest will increase
employment, while beyond this range a rise will reduce it
again. There will then be a certain maximum level of
employment which can be attained by operating upon the
rate of interest, and this maximum may fall short of full
employment. 1 For some communities, therefore, full
employment may lie beyond the reach of even the most
powerful and \nost enlightened of monetary authorities.

In certain communities, on the other hand, which are
poorer, more egalitarian, more spendthrift, more easily
discouraged from saving by a fall in interest, or in which
the elasticity of substitution is low, the monetary authorities
may be in a position to secure full employment in the long-
period by setting the rate of interest at the appropriate
level. It may be contended that we have been guilty of an
over-formalised argument in suggesting that all com-
munities are not of this kind. It is formally true (on our
assumptions) that when the elasticity of substitution is
greater than unity, and the direct effect of a fall in the rate
of interest on the desire to save positive, lowering the rate
of interest causes a decrease in total income and in employ-
ment, but it may be argued that at very low levels of the
rate of interest (as Professor Cassel contended) the desire
to save must be checked, and that at very high levels of
capital per head the elasticity of substitution must fall
below unity. Thus, it may be argued, a sufficient reduction
in the rate of interest will always increase employment.

1 A curve may be drawn connecting the rate of interest with the equilibrium level
of employment, the rate of interest being measured on the.? axis and employment on
the x axis. In the case described above this curve would fall from left to right over the
higher values of j>, reach a point of zero elasticity, and fall back from right to left over
the lower values. This curve may lie over its whole length inside the limits of the
available supply of labour. Full employment is then unobtainable.

2 Upon this view the curve described in the preceding footnote would reach a
second point of zero elasticity and fall from left to right over the lowest values of the
rate of interest. But as the rate of interest falls towards zero a point must be reached
at which the typical earned income becomes greater than the typical capitalist income.
When this point has been passed the condition which formerly obtained is reversed,
and a fall irj the rate of interest tends to increase thriftiness when the elasticity of

92 ESSAYS IN THE THEORY OF EMPLOYMENT

A priori argument on such points is of little value, and it may
very well be that actual investigations would support this
contention.

But even if this view is correct it does not follow that
full employment can always be reached by lowering the
rate of interest. For, first, it may be that no rate short of
zero will be sufficiently low to secure full employment, 1
and second, even if the rate of interest corresponding to
full employment is positive, it may be so- low that it is
impracticable for the monetary authorities to establish it
by any device within their power. 2 It is therefore impos-
sible to maintain that the existence of unemployment is
incompatible with conditions of final equilibrium.

4

Our analysis has so far been conducted upon the
assumption that there are no scarce factors of production.
It will be found that our main conclusions are not affected
by the introduction into our analysis of a fixed supply of
land and other natural resources.

Consider, first, how the existence of a scarce factor
influences the effect upon output of a change in the rate of
interest. In order to simplify the argument we will once

substitution is less than unity. Suppose that we start in a position at which a change
in distribution favourable to capitalists increases thriftiness and at which the elasticity
of substitution is greater than unity. At this point a fall in the rate of interest (assum-
ing that the direct effect upon the desire to save is neutral) will reduce equilibrium
income. As the rate of interest falls towards zero a point must be reached at which
the elasticity of substitution becomes less than unity, and another point must be reached
at which a change in distribution favourable to capitalists (who are now poorer than
earners) reduces thriftiness. When either of these points has been passed, without
the other, a further fall in the rate of interest will raise the equilibrium level of income,
but when both have been passed a fall in the rate of interest, by enriching the earners,
who now enjoy higher incomes than the capitalists, will increase thriftiness and so
reduce the equilibrium level of income. Thus there is likely to be a certain range of
low values of the rate of interest within which a fall will increase equilibrium income,
but in the lowest range of all a fall in the rate of interest will reduce equilibrium
income.

1 The curve may cut the x axis before it has reached the limits of the available
supply of labour.

a General Theory, p. 309.

THE LONG-PERIOD THEORY OF EMPLOYMENT 93

more confine ourselves to the case in which the direct effect
of the rate of interest upon the desire to save is neutral.
An increase in total output will raise the demand for land
and increase the share of landowners in total income.
The extent of the increase in output will therefore depend
upon the saving propensities of the landowners. It seems
natural to assume that in our community the landlords are
similar to the capitalists, both in respect to the range of
incomes which? they receive and in respect to their individ-
ual desire to saVfe. Upon this assumption it can be seen
that the responsiveness of output to changes in the rate of
interest is smaller when land is scarce. If a fall in the rate
of interest increases output by impoverishing capitalists,
it will enrich landlords. 1 And if a fall in the rate of interest
reduces output by enriching capitalists it will impoverish
landlords. The equilibrium total income, with zero
saving, will therefore alter by less, in response to a given
change in interest, when land is scarce than when it is
superabundant. The direction of the change will not be
affected.

The* influence of fixed natural resources upon changes in
the level of real wages must also be considered. When
land is scarce it will no longer be true that a fall in the rate
of interest necessarily raises the rate of real wages. Capital
per head is still likely to be increased by a fall in the rate of
interest, but in those cases in which output expands when
the rate of interest falls, land, per head will be reduced, and
the net effect may be to lower the marginal physical
productivity of labour. In the opposite case, where output
declines as a result of a fall in the rate of interest, land per
head is increased and real wages are raised by so much the
more.

1 This is not quite accurate, for though the increase in output tends to raise land-
lords' incomes, the increase in capital per unit of output consequent upon a fall in
the rate of interest will be partly at the expense of land. There may therefore be a
certain range of outputs over which the landlords are impoverished. But what the
landlords lose by a high elasticity of substitution between land and capital, the
capitalists gain, and on the assumption that both classes have the same degree of
thriftiness, a transference of income between them has no effect upon total output.

94 ESSAYS IN THE THEORY OF EMPLOYMENT

The effect of fixed natural resources upon employment
is in general to reinforce the influence of changes in output.
A given increase in output requires more labour whJn land
is scarce than when it is superabundant and a given
decline causes more unemployment. The case in which
total output increases but employment declines is therefore
less likely to arise when land is scarce.

We have discussed movements in the position of long-
period equilibrium consequent upon changes in the rate
of interest. Other types of change can be similarly treated.
The equilibrium position will be shifted by an alteration
in the thriftiness of the community, whether due to a
change in institutional influences on the distribution of
wealth, to a change in fiscal policy such as an alteration in
methods of supporting the unemployed, to a change in
the degree of monopoly, 1 or to numerous other causes.
And the position of equilibrium will move with move-
ments in population and with changes in technique.

The effect of an increase in thriftiness, as we have seen, 2
is to reduce the equilibrium level of employment and the
stock of capital. The rate of interest being assumed con-
stant, capital per head and real wages are unaffected,
except in so far as land per head is greater at a lower
level of employment. 3

1 In general, an increase in the degree of monopoly will increase thriftiness and so
reduce employment, for it will alter the distribution of income unfavourably to
labour.

Any group of competing firms can increase their aggregate profits, at least tem-
porarily, by amalgamating, and, at any moment, the extent of monopoly is determined
by a tension between the lure of profit, on the one hand, and the desire for indepen-
dence and the difficulties of amalgamation on the other. Since the fear of loss is more
powerful than the hope of gain, there is a strong tendency for the degree of monopoly
to increase as effective demand falls off, while amalgamations formed in slump
conditions tend to break down as trade improves. (Cf. Pigou, Theory of Unemployment,
p. 135). Thus changes in the degree of monopoly tend to amplify the swing of move-
ments in effective demand initiated by other causes. (But see Harrod, The Trade
Cycle, p. 1 7, for a factor which may tell in the opposite direction.)

1 P. 79. * See also p. 112.

THE LONG-PERIOD THEORY OF EMPLOYMENT 95

An increase in numbers produces its effect upon employ-
ment by way of a reduction in thriftiness. If we compare
two positions of equilibrium in which all conditions are
alike except that in one position population is greater than
in the other, then the amount of unemployment correspond-
ing to a given level of employment will be greater in the
first than in the second, and, on any reasonable assump-
tions about the methods by which the unemployed are
supported, the Jevel of consumption will be higher. Thus
the equilibrium tevel of employment in the larger popula-
tion will be greater than in the smaller. 1 With a constant
rate of interest and no scarce factors, the stock of capital
will be increased in proportion to the level of employment
and real wages will be unchanged. If scarce factors are
present, real wages will be lower. Since an increase in
population requires an incfease in capital equipment, to
provide for a higher level of consumption, a continuous
increase in the population would prevent investment from
ever falling to zero.

Changes in technique which result from discoveries of
new methods of production alter the schedule of the mar-
ginal efficiency of capital. If the rate of interest is un-
changed, equilibrium is reached, after inventions have been
made, when the stock of capital is so adjusted that marginal
efficiency, with zero investment, is restored to its former
level. 3

1 It is a simple matter to make a comparison between a community with a larger
and with a smaller population, each stationary. But the process of change, involving
an alteration in the age composition of the population, will set up complicated re-
actions upon thriftiness, through its effect upon the desire and the ability to save
of the representative family; upon productivity, through its effect upon the com-
position of the labour force; and upon the degree of mobility of labour, which in general
is likely to be greater the larger is the proportion of young to old people.

2 The Times is accustomed to point with complacency to the increase in numbers
employed which often occurs even in months when numbers unemployed have also
increased. The implication is that it is a healthy sign that industry can 'find work*
for even a small part of the increased population of working age. But there is no cause
for congratulation in this phenomenon. With every pair of hands God sends a mouth,
and the mere increase of population leads to some increase in employment. For em-
ployment to remain stationary while the population increased would be a sign that
slump conditions were increasing in intensity.

3 I am much indebted to Mr. M. Kalecki for assistance in formulating this analysis.

96 ESSAYS IN THE THEORY OF EMPLOYMENT

Inventions have been classified in various ways. For
our present purpose the most convenient classification
is as follows. 1 A neutral invention is one which affects the
efficiency of production equally at all stages, so that out-
put per head in producing capital equipment is raised
equally with output per head in producing final goods.
When the stock of capital has been adjusted, after an
invention of this type, so as to restore equilibrium to the
given rate of interest, capital per unit of product is the
same as before, and the relative shares of labour and of
capital in a given output are unchanged. A capital-saving
invention increases efficiency in producing capital goods
more than in producing final goods, reduces the equili-
brium amount of capital per unit of product, and reduces
the relative share of capital. A labour-saving, or better
capital-using, invention, increases the equilibrium amount
of capital per unit of output, and increases the relative
share of capital in a given total income.

The effect of inventions upon the equilibrium level of
output will depend upon its reaction on the distribution of
income. An invention which reduces the share of labour
in a given income will reduce the equilibrium level of
output by increasing thriftiness, while an invention which
increases the share of labour will increase equilibrium
output. Thus capital-saving inventions increase, and
capital-using inventions decrease, the equilibrium level of
output.

Equilibrium output will increase or diminish according
as the relative share of labour is increased or reduced.
But even if output is constant, employment will decline,
since the improvement in technique and the increase in
capital per head will combine to increase output per unit
of labour. Thus employment will be constant only if the
situation is such that output increases. Since, in general,

1 Sec The Classification of Inventions. Review of Economic Studies, Februarf, 1938,
where I compare this system of classification with that used by Mr. Hicks in his Theory
of Wages.

.THE LONG-PERIOD THEORY OF EMPLOYMENT 97

capital-using inventions have been the most frequent, 1
there appears to be, from a long-period point of view,
very strong grounds for the popular opinion that inventions
tend to reduce employment. 2

The immediate effect of inventions upon employment
depends upon the extent to which new equipment is the
product of net investment and not merely the result of

1 Mr. Hicks (Theory of Wages, p. 125) distinguishes between autonomous and induced
inventions, the latter be^ng inventions which result from a deliberate effort to think
out ways to take advantage of an alteration in relative factor prices, labour-saving
inventions being induced by a fall in the rate of interest, while autonomous inventions
arise from the progress of pure science or from spontaneous bright ideas. Induced
inventions he subdivides into two types, those whose adoption depends upon the
changed interest rates, and those which would have been profitable in any case, but
which were in fact only thought of under the stimulus of the change. It appears
simpler not to regard the first type as inventions, but rather as adaptations of technique
to changed conditions. When we say that 'technical knowledge is given,' we do not
mean that every entrepreneur has blue-prints in his pigeon-holes of every type of
machine which it would be profitable to use at each conceivable rate of interest.
We merely mean that there is a certain body of technical knowledge which will lead
to a certain adaptation of technique to circumstances. The rapidity and the ingenuity
with which adaptations are made may properly be regarded as a component of the
state of technical knowledge. (Cf. Pigou, Economics of Welfare, p. 217, where a similar
point is made in connection with adaptations resulting from changes in scale.)

Induced inventions in the narrower sense may be of importance. A period of
falling interest rates may be expected to produce a crop of capital-using inventions,
and a period of rising rates, of capital-saving inventions. In such a case the assumption
that the state of technical knowledge is independent of movements of the rate of
interest must be appropriately modified.

Mr. Hicks' opinion that autonomous inventions are likely, over a period of genera-
tions, to be equally divided between labour-saving and capital-saving discoveries,
is hard to accept. It appears obvious that the development of human methods of
production, from the purely hand-to-mouth technique of the apes, has been mainly
in the direction of increasing 'roundaboutness,' and that the discovery of short cuts,
such as wireless, are exceptions to the general line of advance. It is conceivable,
however, that in a community already richly endowed with capital an era of capital-
saving invention may set in.

2 The question has often been posed whether inventions are 'harmful to labour.'
To say that an invention is harmful to labour may mean three things. It may mean
that employment is reduced by it. This, as we see, is highly probable. It may mean
that the real wage rate of those in employment is reduced. This is very unlikely.
A sufficiently labour-saving invention will reduce the marginal productivity of labour
corresponding to the initial amount of capital (cf. Hicks, op. 7., p. 122) but the in-
crease in capital per head required to restore equilibrium tends to raise the marginal
productivity of labour, and it is only in a very extreme case that the real wage rate,
in equilibrium, would be reduced. Finally, an invention harmful to labour may mean
one which reduces the total income of labour. When inventions are neutral the total
income of labour will be unchanged. Therefore, if inventions are mainly labour-
saving, they are likely to* reduce the total income of labour in the long run.

g8 ESSAYS IN THE THEORY OF EMPLOYMENT

using the amortisation funds of old plant to set up new
plant. In general we may suppose that, except when
inventions are highly capital-saving, a period oPpositive
net investment will result from them, even when the
equilibrium stock of capital (reckoned in wage units) is
not increased, for all except the most capital-saving
require an increase in capital per head, while the reduction
in total output which results from increased thriftiness will
not be immediately foreseen. 1 The first feffect of inven-
tions, therefore, is likely to be an increase in employment,
even when in the long run they will reduce it, and a suf-
ficiently rapid succession of inventions, provided they are
not extremely capital-saving, would prevent the rate of
investment from ever falling to zero.

The situation of any actual community can be regarded
as a situation in which investment is tending towards zero.
Before adjustment is reached to a given set of circumstances,
circumstances change. Changes in numbers, in technique,
in the rate of interest, in social and institutional influences
and in the political situation, are constantly shifting the
position of equilibrium, and the processes of investment
never have time to catch up with changes in the equili-
brium stock of capital. The clock is wound up before it
has had time to run down. Even if circumstances remained
unchanged, the system would not run smoothly into an
equilibrium position, for the very process of adjustment
gives rise to oscillations, and even in stationary conditions
the system would fluctuate perhaps eternally, around the
equilibrium position. 2 Moreover, if a position of full long-
period equilibrium were ever reached, with considerable

1 In general, capital-using inventions require a larger amount of investment
than capital-saving inventions, while the amount of investment required to restore
equilibrium with the rate of interest will be greater the greater the elasticity of substitu-
tion. Thus, once more (p. 86), the change most deleterious to employment in the
long period is most beneficial in the short period.

* General Theory , p. 218.

THE LONG-PERIOD THEORY OF EMPLOYMENT 99

unemployment, the nature of the situation would become
obvioii^ to the community, and institutional factors would
be forced to change. Our analysis of long-period equili-r
brium cannot therefore be regarded as a prediction of
the course of history.

Nevertheless it serves to show that long-period influences
are of the utmost importance at any moment of time.
Our static community in full equilibrium is a very remote
abstraction, ancUfor any real community in a changing
world the position of equilibrium is shifting faster than the
system adapts itself to change. But the motive for studying
equilibrium positions is to discover forces which are at
work when the system is out of equilibrium, and the long-
period tendencies which we have discussed are set to work
by any change as soon as, the change is made. When
investment has been going on for a few months new capital
goods come into existence. Labour is displaced by their use;
prices are lowered and the rate of real wages raised as soon
as their product is added to the supply of consumption
goods. .Changes in the distribution of income, raising or
lowering the level of consumption, begin to be felt as soon
as new dividends begin to be paid.

Moreover, the time required for the long-period effects
af a given change to outweigh the short-period effects may
not be very long. Suppose that the effect of a fall in the
rate of interest is considerably to reduce the equilibrium
level of employment. Then when a fall in the rate of
interest has taken place it may be only a matter of a year
;>r two before the level of employment sinks below what it
would have been if the rate of interest had been maintained
at the higher level, in spite of the fact that schemes of
investment made profitable by the fall in the rate of interest
nnay continue to be carried out for many years on end.
Dr, again, since a fall in the rate of interest is likely to
;et up a long-period tendency for real wages to rise, there
/vill be a certain length of time, following an initial fall
n the rate of interest, after which real wages will be higher

loo ESSAYS IN THE THEORY OF EMPLOYMENT

than they would otherwise have been, in spite of the fact
that a high level of employment, tending to depressrurrent
real wage rates, is still being maintained.

In any actual situation long-period tendencies show
themselves in the statistics concurrently with short-period
effects, and it would be impossible to make a comparison
of output, employment and wages between one year
and the next if long-period tendencies were not brought
into account.

THE CONCEPT OF ZERO SAVING

IN conditions of long-period equilibrium, with zero net
investment, net saving, for the community as a whole, is
equal to zero. Long-period equilibrium therefore requires
that the amount and distribution of income and of wealth
shall be such that the total of net saving is zero.

The concept of zero saving is fundamental to the dis-
cussion of long-period equilibrium, equally in the c Classical'
and the 'General 3 system of analysis. In classical conditions
of long-period equilibrium there is full employment, and
the rate of interest is conceived to assume such a value
that full employment is compatible with zero investment.
In the system of analysis outlined in the preceding pages
the rate of interest's taken as given, and the amount of
employment (and total income) is conceived to assume
whatever value is compatible with zero investment at the
given rate of interest. The two systems of thought are
complementary, not rival. For instance, the preceding
analysis has shown that in certain types of society full
employment is not compatible with a positive rate of
interest. The classical analysis, applied to the same
situation, would tell us that, since full employment obtains,
the rate of interest must be negative. 1

It has always been recognised that, even in the classical
scheme, the concept of long-period equilibrium with zero
saving is subject to the difficulty that a fall in the rate of
interest does not necessarily lead to a fall in saving, since
it is associated with an increase in the amount of capital,
andui.may therefore be associated with an increase in

1 A negative rate of interest presupposes the use of some device such as GeselPs
stamped money, for so long as money has no carrying cost there is one form in which
wealth can be held on which the rate of interest is zero, not negative (General Theory,
P 357) W?*en the rate of interest is negative it is profitable to employ capital in forms
which reduce the total output of a given number of men while prolonging the period
of production (General Theory, p. 214).

101

102 ESSAYS IN THE THEORY OF EMPLOYMENT

capitalist incomes. 1 But this difficulty cannot arise unless
the rate of interest is positive. When the rate of interest is
negative an increase in the amount of capital ^neces-
sarily associated with a reduction in capitalist incomes, 2
and, moreover, with a reduction in output per head. There
is, therefore, always some value of the rate of interest,
positive or negative, which will give full employment in
static conditions.

But when the rate of interest is taken as ^iven, and it is
the amount of employment which is conceived to vary,
the problem of zero saving is not so easily disposed of.
The influence of unemployment upon saving may be
separately treated, and we will first assume that the
incomes of capitalists and employed workers are indepen-
dent of the amount of unemployment. 3

Now, the fundamental assumption which is required by
the General Theory of Employment is that an increase in
income leads to an increase in saving. 4 Under short-
period conditions, where an increase in income entails
an increase in employment with given capital equipment,
this assumption is highly plausible, and, in a general way,
the evidence in favour of some measure of stability in the
economic system appears sufficiently strong to warrant
the belief that this assumption is fulfilled in reality. 5 The
corresponding assumption for long-period conditions is
that, the rate of interest being given, the increase in income
which would be associated with an increase in the stock of
capital, when employment of the other factors is adjusted
to it in the proportions dictated by the rate of interest,
would lead to an increase in saving. In short, the existence
of a unique position of long-period equilibrium correspond-
ing to a given rate of interest requires that, if a chance

1 See Knight, Ethics of Competition, p. 183.

* With a negative rate of interest, the elasticity .of substitution is negative, and
consequently less than -f i.

3 For instance, the unemployed may be supposed to fend for themselves by the
methods of 'disguised unemployment'; cf. above p. 6c

4 General Theory, p. 29. 6 Ibid.. D. 2*1

THE CONCEPT OF ZERO SAVING 103

increase in the stock of capital were to occur when equili-
brium has once been reached, then, at the level of income
corresponding to the larger stock of capital, there would
be positive saving. If this condition is fulfilled a chance
increase in the stock of capital will reduce its earnings
below the level dictated by the (constant) rate of interest,
and a period of disinvestment will restore the stock to its
former size. The paradoxical appearance of thinking of an
increase in sa^dng as leading to disinvestment is merely
a reflection of me fundamental paradox that an increase
in thriftiness tends to reduce the stock of capital.

When the above condition is not fulfilled there is no
unique position of long-period equilibrium corresponding
to a given rate of interest. If the desire to save is indepen-
dent of the level of income (given the rate of interest)
then any level of income is' compatible with equilibrium,
and if there is zero saving at one level there will be zero
saving at any other. A chance reduction in the amount of
capital, instead of setting on foot a tendency for the former
stock of capital to be restored, would merely shift the point
of equilibrium to a position compatible with the smaller
stock of capital. From a long-period point of view there
would be neutral equilibrium, though at any moment
short-period stability might still obtain. Further, if a decline
in income associated with a reduction in the amount of
capital leads to an increase in the desire to save, unstable
long-period equilibrium would obtain. Any chance reduc-
tion in the amount of capital would set on foot a tendency
to a progressive reduction in capital and in income.
Such a state of affairs appears uncongenial to common
sense, but cannot be ruled out of court on the evidence,
since evidence relating to static conditions cannot in the
nature of the case be provided by the history of a changing
world.

Complete equilibrium further requires that saving shall
be zero on balance not only for the community as a whole,
but also for each individual. For if some individuals are

104 ESSAYS IN THE THEORY OF EMPLOYMENT

saving, when total saving is zero, others must be dis-
saving, and a progressive redistribution of wealth and of
income will be taking place. The total of capital "is the
same to-day as yesterday, but since yesterday the owner-
ship, of capital has changed hands.

Four cases may be distinguished giving different solu-
tions of the double problem of equilibrium for the system
as a whole and equilibrium for individual savers. First,
it may be assumed that the saving of ea^h individual is
zero over his lifetime. During his youth' 'he is supported
by the expenditure or dis-saving of his parents. In his
active years he saves, and after his retirement he dis-
saves what he has accumulated, providing for his children
in exactly the same manner that his parents provided for
him; In this case there is no progressive accumulation or
decumulation by the individual family. The schedule of
thriftiness for the community as a whole would vary with
the age composition of the population. When the popula-
tion is stable, the dis-saving of the retired exactly absorbs
the savings of the active and total savings are zero at all
levels of total income. A rise in the general Ifevel of
income would lead, after a period of transition, to a
higher rate of both saving and dis-saving, the standard of
life of the retired rising with the standard of the active.

In such a case there is stable equilibrium within the
saving class, in the sense that, if the total of capital is
given, there is no tendency for the distribution of its
ownership to change. But there is neutral equilibrium in
respect to the total amount of capital. Any amount of
capital which happens to be in existence will be preserved
intact, the total of current consumption being always
equal to net income. If there is a chance decline in the
amount of capital, wealth, income, and expenditure will
suffer a corresponding fall and there will be no tendency
for the former stock of capital to be restored. Conversely
for a chance increase in the stock of capital. The assump-
tion that all saving takes the form of simple provision for

THE CONCEPT OF ZERO SAVING 105

old age is therefore incompatible with the assumption
that there is a unique position of long-period equilibrium
corresponding to a given rate of interest.

The second case which we may distinguish arises from
the assumption that the predominant influence upon
saving is the amount of wealth already possessed. In this
case an increase in present income, in itself, may still
increase saving, but an increase in the ownership of capital
will tend to reduce the saving of the capitalist, since his
desire to own An increased amount of wealth becomes
weaker as the amount he owns increases. Let us suppose
that an individual, the whole of whose income is unearned,
saves less when his income and his wealth increase in the
same proportion (as would occur when the rate of interest
is constant), and more when his income and wealth decline.
On this assumption there is an equilibrium distribution
of capital between individual owners corresponding to a
given total of capital. Suppose that the total of capital
is given. Then if equilibrium within the capitalist class
does not obtain, some individuals, who own little capital
or who are thrifty by nature, will be saving, and others,
who are more wealthy or more spendthrift, will be dis-
saving. But as capital changes hands, those who gain
capital become less anxious to acquire more, and those
who lose become more anxious to avoid a further loss.
The rate of saving of the one class, and dis-saving of the
other, fall off as capital changes hands, and equilibrium
is reached when the ownership of wealth is adjusted to
temperament, the most miserly being the most wealthy, in
such a way that no individual is either saving or dis-saving.

In this case equilibrium within the capitalist class can
be attained. But the system as a whole is in unstable
equilibrium. For in this case a chance increase in the
amount of capital (the rate of interest being constant)
would reduce the desire to save of those individuals to
whom the new capital accrued, Thriftiness would be
reduced and a progressive increase in the stock of capital

106 ESSAYS IN THE THEORY OF EMPLOYMENT

would begin. Conversely, a chance reduction in the
amount of capital would increase thriftiness and a progres-
sive decline in the amount of capital would set irv In so
far as earned incomes are partly saved a stabilising factor
is introduced into the picture. For a chance increase in
capital would raise the equilibrium level of earned income,
and the income of the earning classes would increase in a
greater proportion than their wealth. For this class,
therefore, saving may increase with an increase in the
total of capital. But if the bulk of saving is done by
individuals the whole of whose income is unearned, then,
in the case which we are considering, there is no equili-
brium value of the total amount of capital corresponding
to a given rate of interest. Thus this assumption, which
gives stable equilibrium within the capitalist class, gives
unstable equilibrium for the system as a whole.

Neither of these two cases has much appearance of
reality. It is natural to suppose that an increase in income
would lead to an increase in the habit of bequeathing
wealth to descendants, so that the first case could not
occur. The second case appears equally unreal, for it is
natural to suppose that the amount of wealth which an
individual requires to provide a sense of security rises with
his standard of life, and that the amount which he requires
to satisfy pride rises with tljie amount possessed by his
rival capitalists, so that a general increase in the income
of the capitalist class, even though it is associated with a
proportionate increase in their wealth, leads to a rise,
not a fall, in the amount which they desire to save. This
view is congenial to common sense, since a richer com-
munity does not normally exhibit a lower propensity to
save than a poorer community. It is necessary to admit
the theoretical possibility that the assumption that saving
increases with income may fail to be fulfilled under long-
period conditions, and when it is not fulfilled there is no
unique position of long-period equilibrium corresponding
to a given rate of interest. But to pursue the analysis of

THE CONCEPT OF ZERO SAVING 107

neutral or unstable long-period equilibrium would involve
an expenditure of ingenuity which scarcely seems profit-
able. It is better boldly to postulate, what is after all
highly plausible, that an increase in total income would
lead to an increase in saving, even when it is associated
with an increase in wealth. The foregoing analysis of
the long-period theory of employment was conducted
upon this assumption. Upon this assumption our first
two cases are ruled out of court.

There is a^hird case, in which stable equilibrium
obtains both for the system as a whole and within the
capitalist class. In this case it is assumed that an increase
in the general level of wealth does not discourage saving,
but that an increase in relative wealth does. Then to each
amount of total capital will correspond a certain distribu-
tion between individual capitalists which will induce zero
saving for each individual, as in our second case. The
pride and ambition of the savers will reach saturation as
their wealth increases relatively to the average of the
capitalist class, while the humiliation of losing will after
a point induce the dis-savers to live within their means.
But in this case a chance increase in the total of capital
would not reduce the motive for saving, since the average
ownership of wealth, which is assumed to govern the desire
for possession, would rise with the total. If the new
capital is imagined to accrue to a particular group of
capitalists, they would be induced to dis-save, but envy
will now inspire saving on the part of others, and after
a process of re-shuffling an equilibrium distribution of
ownership will be attained. The level of total income
corresponding to zero total saving will be lowered when the
average ownership of capital is increased, and there will
be a unique equilibrium amount of capital corresponding
to a given rate of interest. Thus a position of stable
equilibrium exists for the system as a whole as well as for
the capitalist class.

A fourth case .arises from the assumption that the saving

io8 ESSAYS IN THE THEORY OF EMPLOYMENT

of an individual increases with his present income, even
though his wealth has increased at the same time. In
this case there is stable equilibrium for the system* as a
whole, since a chance increase in capital would lead under
long-rperiod conditions to an increase in saving. But there is
no equilibrium within the capitalist class. Any individual
who happens to save will augment his income and so
increase his rate of saving. While any individual who is
induced to dis-save by a decline in income will reduce his
income progressively and increase his ens-saving, until
the whole of his wealth has disappeared and he is driven
from the ranks of the capitalist class. In this case, the
total of capital being given, there is a progressive concen-
tration of wealth going on at any moment, of which the
only logical resting-place is reached when the whole
national capital is owned by the most pertinacious saver
in the capitalist class. Bizarre as it may appear, this case
is in some respects the most realistic of the four which we
have considered. For it may very well be that in a modern
capitalist society there is a strong underlying tendency
towards a progressive concentration of wealth, which is
only held in check by periodic redistribution through
accidents, death duties, the division of large estates
between several heirs and the occasional appearance of a
spendthrift in a wealthy family.

The underlying tendency towards concentration may
be partially offset by a decline, as time goes by, in the
thriftiness of wealthy families. The sons of Marshall's
successful business man 'are perhaps left a good deal to
the care of domestic servants whfl are not of the same
strong fibre as the parents by whose influence he was
educated, 5 and instead of concentrating on business
ambition they are 'at least equally anxious for social or
academic distinction 3 ,* so that the family firm withers
like a tree in the forest; similarly the sons of the assiduous
saver may turn from their father's ways, and the family

1 Principles, p. 300.

THE CONCEPT OF ZERO SAVING 109

fortune begin to decay. Equilibrium of the Marshallian
type .would be established when there was zero saving on
balance within each grade of the capitalist class, while
the personnel of each class was altering. At any moment
some individuals within a grade would be saving and rising
out of it, and some dis-saving and falling below it, their
places being taken by dis-savers from a higher grade
and savers from a lower grade. There would then be no
progressive teixjency towards the concentration of capital
in ever fewer hands.

Even when the tendency to concentration prevails over
the tendencies which bring about redistribution, so that
there is no equilibrium within the capitalist class, there is
still, as we have seen, a unique equilibrium total amount
of capital corresponding to t a given rate of interest at any
moment. But if concentration is proceeding through time,
so that the distribution of income is becoming progressively
more unequal, then the thriftiness of the community will
be increasing, and the equilibrium amount of capital,
corresponding to a given rate of interest, will fall as time
goes by. Thus as concentration proceeds the equilibrium
level of income and employment will fall. The final posi-
tion of equilibrium is not reached until the whole capital
of the community is owned by one man, and his capital
is reduced to the point at which his savings are zero.
The evils of unequal distribution will then have produced
their maximum effect.

The paradoxical appearance of the foregoing discussion
is partly to be attributed to the fact that we had ruled out
of account an important influence tending towards
stability. We have yet to examine the effect of unemploy-
ment upon the thriftiness of the community. A reduction
in the net income of capitalists and employed workers
due to the fact that they are obliged to support the unem-
ployed may be assumed to reduce the amount that they
save. Thus, in general, the larger the amount of unemploy-
ment corresponding to a given level of employment, the

i io ESSAYS IN THE THEORY OF EMPLOYMENT

smaller will be the amount of saving corresponding to a
given level of employment, and the higher will be the
equilibrium level of employment. Moreover, the 'equilib-
rium level of employment will tend to be higher the more
lavish is the provision made for the unemployed.

Conditions of neutral or unstable long-period equilib-
rium are less likely to arise when the unemployed are
supported by the rest of the community than when they
are left to fend for themselves. For an increase in the stock
of capital, with a given rate of interest, would be accom-
panied, if it persisted in equilibrium conditions, by an
increase in employment. But a reduction in the drain on
the incomes of the rest of the community caused by un-
employment would lead to an increase in saving. Thus
the condition for stable equilibrium, that a chance increase
in capital leads, under long-period conditions, to an
increase in saving, is more likely to be fulfilled the more
ample is the provision for the unemployed.

So long as the unemployed are supported at the expense
of the rest of the community, that is to say by taxation or by
charity, there is no disturbance to equilibrium within the
capitalist class. But we must also consider the case in
which the unemployed are supported by borrowing, either
privately or by the state.

In this case the incomes of the rest of the community,
when the system as a whole is in equilibrium, will be just
so great as to induce a rate of positive saving equal to the
negative saving represented by borrowing on behalf of
the unemployed. Equilibrium within the capitalist class
will not then exist, for accumulation must continue as
long as borrowing persists. If to-day inherits yesterday's
debts, to-day is not a simple repetition of yesterday.

A source of disturbance is now introduced into the
equilibrium situation. For, when the amount of unem-
ployment is considerable, the ever-mounting burden of
debt must sooner or later reach a point at which it becomes

THE CONCEPT OF ZERO SAVING in

intolerable, so that some means have to be found to curb
the exactions of creditors. In such a case the institution of a
Jubilee Year, at which all debts are forgiven, might fend
off political upheavals and enable the community to con-
tinue indefinitely supporting the burden of long-period
unemployment.

DISINVESTMENT

WHEN the stock of capital is in equilibrium a rise in the
rate of interest, or a fall in prospective earnings, will set
up a tendency for disinvestment to occur. But the extent
of the disinvestment will be different for different types of
capital goods. ,

In this connection capital goods may be divided into
two classes, according to the method by which their effi-
ciency is maintained. A machine of one type has a certain
period of working life, after which it is discarded and
replaced by a similar new machine. The other type, of
which a railway bridge is an example, has, practically
speaking, permanent life, provided that running repairs
are carried out. In many cases either method of main-
tenance is technically possible, but one will be less costly
than another. A building could be left without adequate
repairs until it began to fall to pieces, and another building
could then be erected in its place, but if this method were
pursued the amortisation fund required permanently to
maintain the equivalent of the original building would be
a greater proportion of its initial cost than if the method
of continuous repairs was adopted. A lick of paint from
time to time will save expense in the long run.

The extreme case of the first type is to be found in, for
example, a stock of coal, of which the whole initial cost
(assuming prices are unchanged) has to be laid out each
time the capital is used if the initial stock is to be main-
tained intact. An extreme example of the second type is
the preparatory survey for a railway line, where the capital
is maintained permanently intact without any expenditure
above the initial cost. From the point of view of the genera-
tion that inherits it, capital of this type is equivalent to
"the free gifts of nature.' A border-line case is provided
by a machine which has, say, ten years 'of working life at

112

DISINVESTMENT 113

full efficiency if no repairs are undertaken, and which has a
permanent life if 10 per cent, of its initial cost is laid out
each year on repairs. In this case an amortisation fund of
10 per cent, of cost is required to maintain efficiency by
either method, and it will be a matter of indifference
(neglecting interest) which method is pursued. 1

The risk involved in the initial investment in equip-
ment of the second type is greater than in the investment
of an equal sum in the first type, Againsi the first type an
amortisation fuild is accumulated during the life of the
machine, and if, when the time for renewal comes, the
prospects of that type of machine are less bright than they
were when it was first made, the fund can be used for
buying a different machine or for holding a security,
instead of replacing the defunct equipment. But where the
method of repairs is used eath year's contribution to amor-
tisation is sunk in the equipment as it is made and cannot
be used for acquiring any type of asset other than the
machine in question. If two pieces of equipment, one
of either type, each with an amortisation fund of 10 per
cent. 6f cost, both lose all prospect of future earnings
in the fifth year of life, the owner of the machine of the first
type can save half the cost of the machine from the wreck-
age, while the owner of the other machine can save nothing.
With equipment of the first type the possibility of c getting
your capital out of the business 5 in case of need is much
greater than in the second type. If the most probable
prospective profits are equal in the two cases when the
initial investment is made, the degree of certainty must be
greater to induce an investment of the second type, or,

1 Many cases will present a more complex problem. In general, the efficiency of a
machine will fall off gradually, if no outlay is made on repairs, until it finally becomes
useless. The more is spent upon repairs the greater will be the efficiency of the machine
for a given length of life, and the greater will be the length of life at a given level of
efficiency. It is therefore necessary to balance the cost of repairs against the decline
in earnings of the machine which will occur, in a given state of the market, if repairs
are not undertaken. In most cases where the replacement method is used some
current repairs will also be profitable. Equipment belongs to the second type when it is
profitable to maintain perpetual life at a given level of efficiency by means of repairs
alone.

u 4 ESSAYS IN THE THEORY OF EMPLOYMENT

with a given degree of certainty, prospective profits must
be higher. 1

For equipment of the first type the process of dis-
investment is perfectly symmetrical with the process of
investment. But with equipment of the second type the
case is different. If the rate of interest rises after the
investment has been made the machine will continue to be
maintained so long as its current yield exceeds its current

upkeep.

s

A simplified example will make the poirrt clear. Suppose
that there is a set often similar machines each costing 100
to produce, with a length of life of ten years, and suppose
that at a particular moment the age of the machines is
evenly distributed, one dating from each of the past ten
years. Suppose that the expected earnings of the set of
machines are 150 per annum. Then the outlay of 100
to replace the oldest machine when it wears out is equiva-
lent to the purchase of a perpetual annuity of 5 (the
earnings of the machine being 15 and its depreciation
allowance 10 per year). 2 Thus if the rate of interest is
5 per cent, the machine will be replaced. If the 1 rate of
interest is greater than 5 per cent, it will not be replaced.
At the end of a year, unless prospective earnings have risen
or replacement cost fallen, a second machine will be
allowed to wear out and so forth, the stock of machines
gradually dwindling, and being replaced, from the point
of view of their owner, by securities. Now suppose that a
single machine, costing 1000 to produce, can be kept in
repair by an annual outlay of 100, and suppose that if
this outlay is not made in any one year the machine

1 In the border-line case given above the method of renewal will for this reason be
preferred to the method of repairs. In a chancy market the border line will be pushed
further into the territory of the method of repairs than would be the case if future
prospects were known with certainty. Risk is allowed for in the calculation of pros-
pective profits, and for simplicity two pieces of equipment may be described as having
equal prospective profits if these prospects are equal after allowance has been made for
any difference in risk.

2 In this illustration, as in that above, interest earned on the amortisation fund is
left out of account in order to simplify the calculation.

DISINVESTMENT 115

becomes unusable. Suppose as before that its earnings are
150 per year. Now an outlay on repairs of 100 in
any one <y ear results in a profit of 50. Although it would
not pay to produce the machine in the first place if the
rate of interest were more than 5 per cent., it will not pay,
once it exists, to let it fall out of repair, whatever the rate
of interest may be. The continued existence of the plant
after the rate of interest has been raised will prevent pros-
pective profits . from being tradually increased by the
gradual disappearance of similar machines, and the price
of the machine will remain permanently below its cost of
reproduction.

Thus if the rate of interest falls below 5 per cent., the
stock of both types of equipment will be increased, but if it
rises above 5 per cent, the stock of the first type only will
be reduced. 1 Similarly a small fall in prospective profits
will lead to disinvestment in the first but not in the second
type of equipment. 2

This example is highly simplified. Normally the life
of a machine will not come to an abrupt end as soon as
expenditure on repairs is cut down, and a rise in th rate
of interest, or decline in prospective earnings, would lead
to some reduction in the amount of repairs that it is
profitable to carry out. But in the normal case the distinc-
tion between the two types of equipment, though less
dramatic than in this example, will still be present.

If a significant amount of plant is of the type which is
maintained by repairs, it follows that the amount of
capital per head in a position of equilibrium is not uniquely
determined by the rate of interest. Any period of invest-

1 Cf. the distinction between capital goods and *rent-earning goods' drawn by Wick-
sell, Interest and Prices, p. 1 26. A similar point is discussed in connection with the notion
of normal profits by Mr. G. F. Shove, Economic Journal, March 1933, p. 119; by Mr.
R. F. Harrod, Economic Journal, June 1933, p. 337 and Quarterly Journal of Economics,
May 1934, p. 457; and by myself, Quarterly Journal of Economics, November 1934,
p. 108.

1 A rise in the cost of capital goods, relatively to their yield, has an equal effect on
both types, provided that the rise affects repairs and replacement in the same propor-
tion.

u6 ESSAYS IN THE THEORY OF EMPLOYMENT

ment will leave a permanent legacy behind it, and if a
given position is reached after a period of low interest
rates, or exaggerated expectations of profit, capital per
head will be greater,' real wages higher and the level of
profits lower, than if the same position had been reached
with a past history of lower rates of capital accumulation.

This consideration is of great importance in many
departments of economic analysis. Its most interesting
application is to the problem of declining population.
If a given level of population is reached after a period of
decline, its capital equipment will be greater than if the
same population had existed for some time in a condition
of stability or were reached after a period of expansion.
In a community where population has declined there
will be a large amount of plant whose price is less than
its cost of production. The earnings of this equipment
have fallen below the level which was necessary to tempt
it into existence in the first place, but are not yet so low
as to cause it to fall out of use. Real wages are conse-
quently higher than would otherwise be the case. In effect
certain capitalists have been disappointed into providing
capital more cheaply than they intended, and the rest
of the community is benefited at their expense.

Now it happens that the class of capital whose prospec-
tive earnings are most affected by changes of population
houses, land drainage, roads and railways is also a class
in which the repair method of maintenance is predominant.
Thus the principle governing disinvestment is of the great-
est importance to a community with declining population.
A smaller population inherits housing and permanent-
way designed for its more numerous ancestors. Rents and
fares are low. The railway companies pay small dividends,
and at some stage may have passed through a process of
bankruptcy, ruining their debenture-holders but enriching
the rest of the community.

Let us compare two communities, ea^h of which is in

DISINVESTMENT 117

full equilibrium with a stationary population, 1 which are
alike in every respect except that in one, say Alpha, capital
is predominantly of the type which is maintained by
repairs, while in the other, Beta, it is maintained by
renewals. Since, in the extreme case, capital of the first
type is equivalent to natural resources, we may suppose
that in Beta land is superabundant. In face of a decline
in population, each community will suffer a decline in
employment. 2 ^hc reduction in consumption, expected
to be permanent, will lead to a decline in the stock of
capital. The decline in employment will be greater in
Beta, not only while disinvestment is proceeding as a
result of the reduced profitability of capital goods, but also
permanently, since in Alpha thriftiness is likely to be
permanently reduced by the impoverishment of capitalists.
In each there will be an initial rise in real wages, for with
reduced employment, capital per unit of labour will
increase. 3 But in Beta the stock of capital will decline as
time goes by and real wages will gradually relapse towards
their former level, while in Alpha the predominant part
of the stock of capital will remain in existence and real
wages will be permanently raised.

Thus it appears that (except so far as the capitalists are
concerned) Alpha is in a more favourable position than
Beta, in face of a decline in population. But against this
must be set a serious disadvantage. If any circumstance
favourable to investment, such as a decline in the rate of
interest, were now to occur, the consequent increase in
employment will be much greater in Beta than in Alpha.
Let us suppose that, after the decline in population has
occurred, the price of every kind of capital good in Alpha
is less than its replacement cost. A small fall in the rate of

1 The argument can readily be generalised to cover the case in which long-period
equilibrium does not obtain in the initial position.

2 See p. 95.

8 The notion of an 'optimum population* at which average real income is maximised
can only be discussed in long-period terms, for, from a short-period point of view,
population is always greJter than the optimum.

n8 ESSAYS IN THE THEORY OF EMPLOYMENT

interest will then lead to a rise in the price of capital goods
in Alpha, but to no other effect, whereas in Beta it will
inaugurate a period of investment and increased employ-
ment. After the lower rate of interest has ruled for some
time the stock of capital in Beta will have increased
and the difference between the two communities will be
lessened. But Beta will enjoy the short-period benefits
of investment, from which Alpha is debarred.

This is upon tke assumption that the frep play of private
enterprise is allowed to determine the level of employment.
It would always be possible for the authorities to offset
the deleterious effect of a decline in population upon
investment by taking measures calculated to increase the
propensity to consume.

DIAGRAMMATIC ILLUSTRATIONS

AN attempt to submit the General Theory of Employment
to diagrammatic treatment is subject to grave objections,
since few of the concepts concerned can be reduced to
precise quantitative terms. A formalised treatment may,
however, be o use merely as an alternative method of
expressing some of the propositions of the foregoing
analysis. It will be found necessary to make a number of
simplifications in order to reduce them to a form which
can be expressed in a two-dimensional scheme.

THE SUPPLY CURVE OF LABOUR

It is clearly permissible to say that, in some sense,
the amount of labour which a given population 1 is able
and willing to provide will vary with the ruling level of
real w^ges. But the conception of a given amount of
labour presents certain difficulties. First of all the notion
of the total supply of labour can be given no precise
meaning for a system within which labour is immobile,
between areas or between trades, and our illustrations

1 An attempt is sometimes made to treat movements of population as a function
of real wages. In the simple Malthusian view the supply of workers is perfectly
elastic at a certain wage, representing the minimum standard of life. But even in those
countries where Malthusian conditions prevail, the decline in the death-rate which
follows an improvement in real wages does not come about in any simple or immediate
way, but shows itself partly in, for instance, a better resistance to the next epidemic
which happens to follow the improvement, and partly in the enhanced efficiency of
those who survive. Professor Pigou (Stationary States, p. 167) restates the Maithusian
argument in terms not of a physical minimum but of a customary standard of life.
On this view, if, for instance, agricultural workers are used to enjoying wireless sets,
we must expect that they will refuse to be born when wages fall to a level at which
they can no longer afford them. It is sometimes argued that in the Western world the
Malthusian law is reversed (cf. Robertson, Economic Fragments, p. 16) and that a
community which has become accustomed to an improved standard of life will exper-
ience a decline in the birth rate. But neither the Malthusian nor the reverse- Malthu-
sian law works with sufficient regularity for the notion of population movements as a
function of real wages a to be a profitable simplification, and it seems preferable to
treat population movements as an independent datum.

120 ESSAYS IN THE TffEORY OF EMPLOYMENT

must be confined to cases in which either there is a high
degree of mobility or the demand for labour of various
types always moves in such a way as to require no change
in their relative supply. When this difficulty is disposed
of, others remain. The amount of work which an individual
provides in a day varies with his personal efficiency, with
the intensity of his efforts and with the number of hours
worked per day, none of these variables being completely
independent of tbe others, and the number, of worker-days
which a given population provides varies with the average
number of workers per family. 1 Moreover, a given pro-
portional increase in one dimension of the supply of work
will not produce the same effect upon the total, along
whichever dimension it is made. To double the number
of workers, keeping hours unchanged, is not equivalent
to doubling the number of hours, keeping the number of
workers the same. A unit of work is therefore a conception
of some complexity. Similarly, the level of real wages is
not a simple rate of earnings per unit of work but a highly
complicated system of interrelated rates. In any particular
inquiry the relevant complexities must be taken into
account, but for many problems a rough-and-ready treat-
ment will suffice. For the purely illustrative purpose of
the present note we will suppose that circumstances are
such that it is possible to construct an appropriate com-
pensated unit of work, representing a man-day of given
efficiency, and an appropriate index of real wage rates
per unit of work. It is then possible to draw a curve
connecting the amount of labour which the population
wishes to provide with the ruling rate of real wages.

A rise in the level of real wages will have effects which
tell in opposite directions. On the one hand it will raise
the personal efficiency of the workers by improving their
standard of life; on the other hand, as needs become less
urgent, the choice between an increase in leisure and an
increase in earnings is likely gradually to turn more and

1 Robertson, he. cit., p. 7.

DIAGRAMMATIC ILLUSTRATIONS 121

more in favour of leisure, in spite of the extra inducement
to work represented by a higher rate of reward. As the
rate of Eeal wages rises, hours of work will generally tend
to be reduced, the working life of the representative
individual will be shortened both by longer education
and earlier retirement, and women will be less anxious to
find employment. 1 Though the question is not entirely
beyond dispute, it may be regarded as a general rule that
the amount of work provided by a giveii family of given
efficiency will fall as real wage rates rise. 2 Thus as the level
of real wages rises a decrease in hours of work and in the
number of workers in the representative family must be
set against the improvement in personal efficiency. At
very low levels the latter must be of preponderating
importance, and when the ^standard of life falls below a
certain point the supply of labour must contract even
though every available person is prepared to work to the
maximum physical limit. As the level of real wages rises
the improvement in efficiency due to an increment of
earnings will become less and less, and a point will be
reached at which the effect of reduced willingness to work
begins to outbalance the effect of increased ability. A

1 The choice between earnings and leisure is not, in modern conditions, left entirely,
or even mainly, to the preference of the individual, but is standardised by collective
decisions, legal or customary, as to the length of the working day, the period to be
spent in education, and so forth. These decisions, however, will be largely influenced
by the urgency of the needs which have to be sacrificed to secure the benefits of
leisure (cf. p. 49).

2 Marshall (Principles, p. 528-9) held that this generalisation only applies to those
who live in 'southern climes'. While 'those whose mental horizon is wider, and who
have more firmness and elasticity of character, will work harder and longer the higher
the rate of pay which is open to them.' And he concludes 'that increased remunera-
tion causes an immediately increase in the supply of efficient work, as a rule; and that
the exceptions to this rule . . . are seldom on a large scale.' Weight must be allowed
to this view, but there is a considerable body of evidence against it (see, e.g., Douglas,
Theory of Wages, chapters xi and xii). Mr. Harrod's attempt to disprove it by a priori
methods (Economic Journal, December 1930, p. 704) unfortunately breaks down, for
it is not the case, as he asserts, that a curve, which has less than unit elasticity at both
ends, cannot (apart from 'kinks and irregularities') have more than unit elasticity over
an intermediate range.

A rise in the marginal above the average rate of pay, such as is offered under over-
time schemes, will norm^Jly call forth additional work. Such schemes introduce some
complications which are not here discussed (cf. Robertson, loc. t&, p. 8).

122

ESSAYS IN THE THEORY OF EMPLOYMENT

curve representing the supply of labour will therefore be
of some such form as that shown in Figure i, at first
forward rising, reaching a point of zero elasticity and then
becoming backward-rising.

f

Amount of labour x

FIG. I

The supply of labour reckoned in units of work is
measured on the x axis and the level of real wages
per unit of work on thejy axis in this and the following
diagrams.

This curve is of use solely in depicting the point of full
employment. It does not represent the amount of work
which will actually be done at any level of real wages,
but the amount of work which the population is willing
to provide.

In a case in which either Trade Union policy or spon-
taneous feeling dictate a minimum level of real wages,
such that workers will prefer to face the hardships of un-
employment rather than accept it (p. 28, n. i), the lower
part of the curve will be highly elastic at a rate of real
wages corresponding to this minimum. In face of a lasting
period of high unemployment the minimum level of real
wages demanded is likely to fall, for though wage rates
tend to rise as unemployment increases the average earn-
ings of the whole working population will be low when
unemployment is high, and both the customary standard

THE FOREIGN EXCHANGES 123

of life and ability to hold out for terms will be declining. 1
The level of unemployment experienced over a number
of years will affect the supply curve of labour in other ways.
On the one hand a high level of unemployment, just as
much as a low level of real wages, may lead to a loss of
efficiency and so cause the available supply of labour to
fall off. On the other hand hours and intensity of work are
increased not only by a fall in real wages but also by an
increase in insecurity, which causes workers to placate
their employers, in fear of losing their jobs, by working
hard and refraining from complaints about unpleasant
conditions. When the actual level of employment is high
workers become more exacting and . are less willing to
endure long hours or to submit to being speeded up.
Thus a high level of unemployment will tend to reduce
the available supply of labour in one way and increase it
in another. The net effect of these changes may be repre-
sented by a shift in the supply curve. But such considera-
tions show one of the imperfections of the method of treating
the available supply of labour in a given population as a
function of real wages alone.

THE LONG-PERIOD DEMAND CURVE FOR LABOUR

The relation between real wages and the actual amount
of employment must next be considered. A curve can be
drawn up, in given conditions, relating the amount of
employment to the level of real wages which would rule
if that amount of employment were to obtain under
equilibrium conditions. This curve will not be indepen-
dent of the manner in which the amount of employment is
conceived to alter, and a different curve must be drawn up
for each type of movement. If a change in employment is

1 The minimum level will be conceived in terms of wages per man, whilst our
curve is drawn up in terms of wages per unit of work. The weakening of resistance,
may take the form of willingness to work longer hours for the same daily wage. The
paradoxical expedient of lengthening the working day in the face of severe unemploy-
ment is not unknown in practice.

i2 4 ESSAYS IN THE THEORY OF EMPLOYMENT

conceived to be due to a movement of the rate of interest
it will be accompanied, in the long period, by a change
in capital per head, and consequently in the marginal
physical productivity of labour. If it is conceived to be due
to a change in thriftiness, with a given rate of interest,
capital per head will be unaltered (except in so far as a
change in land per head induces a change in capital per
head), but the relative supply prices of wage-goods and
non- wage-goods dre likely to be affected. This curve has
some affinities with the conception of a demand curve,
since it relates the level of employment to the correspond-
ing wage rate. But it is fundamentally different in nature
from an ordinary curve. The rate of wages is not an
independent, and the amount of employment the dependent
variable. Both are dependent upon variations in the rate
of interest or the level of thriftiness. If circumstances are
such that the level of employment is x 9 then the same cir-
cumstances produce a real wage rate^. For lack of a better
term the curve will be described as a demand curve for
labour, but it is important to bear in mind the distinction
between this curve and an ordinary demand curve.

To illustrate some of the propositions set out in the fore-
going discussion of the 'Long-period Theory of Employ-
ment' we will make use of the curve appropriate to changes
in the rate of interest. If technical conditions, natural
resources and the degree of monopoly are given, and if
thriftiness is unchanged except in so far as it is a function
of the rate of interest, there is a unique level of employment
corresponding under conditions of long-period equilibrium
to each rate of interest and a unique level of real wages
corresponding to each level of employment. A demand
curve can then be drawn up depicting the change in the
level of real wages which accompanies a change in the
equilibrium level of employment brought about by an
alteration in the rate of interest. We have seen that a
fall in the rate of interest may either increase or decrease
the equilibrium amount of employment (pp. 81-86).

DIAGRAMMATIC ILLUSTRATIONS

125

But, when there are no scarce factors, a fall in the rate of
interest always raises the level of real wages. Thus the
elasticity of our curve may be either positive or negative.
When scarce factors are present (p. 93) a fall in the rate of

FIG, 2

FIG. 3

interest which leads to an increase in employment will be
accompanied by a smaller rise in real wages than if there
were no scarce factors, or even by a fall, while a fall in
the rate of interest which reduces employment will be
accompanied by a greater rise in real wages. Thus the
effect of scarce factors is to increase the elasticity of the
curve.

Figures 2 and 3 represent two of the many possible forms
which may be assumed by the long-period demand curve
appropriate to changes in the rate of interest, the direction
of the arrows showing the direction of a fall in the rate of
interest. Figure 2 depicts a case in which scarce factors are
unimportant, so that a fall in the rate of interest always
raises real wages, and in which successive reductions in the
rate of interest at first increase and then reduce the equili-
brium level of employment (p. 89). Figure 3 depicts a case
in which a fall in the rate of interest always increases
employment, and in which the effect of scarce factors
becomes increasingly important as employment expands,
so that successive reductions in the rate of interest at first
raise and then lower the equilibrium rate of real wages.

126

ESSAYS IN THE THEORY OF EMPLOYMENT

FULL EMPLOYMENT

The point of full employment is defined by the point of
intersection of the supply curve and the demand curve.
When effective demand is such that actual employment
stands at this level, then every unit of work available at
the corresponding level of real wages is bespoken and one
entrepreneur can increase the amount of labour he em-
ploys only by reducing the amount employed by someone
else. Figures 4, 5, 6, and 7 represent cases in which the

w

FIG. 5

FIG. 6

FIG. 7

D in each diagram is the demand curve, and S the
supply curve. OF represents full employment and
OW the real wage rate which rules when full employ-
ment is attained.

elasticities of the demand and supply curves have the same
or different signs at the point of intersection,

DIAGRAMMATIC ILLUSTRATIONS 127

Beyond the point of intersection the demand curve lies
below the supply curve when the supply curve is rising
and above it when it is backward-rising. 1 For, where the
curves cut in the contrary sense an increase in effective
demand, at the point of intersection, is accompanied by
a change in the level of real wages which calls forth a
greater increase in the available supply of labour, and full
employment does not obtain. This rule applies equally
whether a rise or a fall in real wages is associated with an
increase in employment, and equally whether a fall is due
to the existence of scarce factors or to the fact that an
increase in equilibrium employment requires a rise in the
rate of interest. The direction of the arrows shown in
Figures 2 and 3 is therefore irrelevant in the present
context.

In some circumstances no 'point of intersection exists
between the supply curve and the long-period demand
curve, because the long-period demand curve appropriate
to changes in the rate of interest lies entirely within the
supply curve (p. 90) as in Figure 8. But reductions in
thriftiness with a given rate of interest lead to no increase in
output per head. The demand curve appropriate to changes
in employment brought about by changes in thriftiness will
have infinite or negative elasticity (as in Figure 9), and since
there must always be some level of employment at which
the available supply of labour becomes perfectly inelastic,
a point of intersection between the two curves must always
exist. In short, while there may be circumstances in which
full employment cannot be reached by manipulation of

1 Mr. Keynes' definition of involuntary unemployment is not strictly accurate.
'Men are involuntarily unemployed if, in the event of a small rise in the price of
wage-goods relatively to the money-wage, both the aggregate supply of labour
willing to work for the current money-wage and the aggregate demand for it at that
wage would be greater than the existing volume of employment* (General Theory,
p. 15). This condition is fulfilled by the level of employment OF in Figure 7 at which
there is no involuntary unemployment. Mr. Keynes' definition is appropriate only
to short-period conditions, when a fall in real wages is a symptom of an increase in
effective demand, and it is formally correct only for cases in which the elasticity of
supply of labour in terms ,pf real wages is zero or positive.

128

ESSAYS IN THE THEORY OF EMPLOYMENT

the rate of interest, it can always be reached by a sufficient
reduction in thriftiness (p. go).

D

FIG. 8

FIG. 9

It is important to bear in mind that the point of inter-
section of these curves is not of the same nature as the point
of intersection of ordinary supply and demand curves.
It does not represent a point of equilibrium, but merely the
point beyond which it is impossible for employment to
expand.

THE SHORT-PERIOD DEMAND CURVE

Any given point on the long-period demand curve is
intersected by a short-period demand curve, showing the
course that real wages would follow if a change in employ-
ment were to occur after equilibrium had been reached
with the stock of capital appropriate to that point on the
long-period curve. The short-period curve falls off steeply
for an increase in employment above the kmg-period
equilibrium amount, for with given equipment the mar-
ginal physical productivity of labour falls with an increase
in employment; while it rises above the long-period curve
for a decline in employment, which reduces the intensity
with which the given stock of capital goods is employed.
This curve also must be drawn with reference to the
manner in which a change in effective demand is brought
about. In the case illustrated in Figure 10 changes are

DIAGRAMMATIC ILLUSTRATIONS

129

Fia. 10

Dl is a long-period demand curve for labour.

DSi is the short-period curve appropriate to the point

P! and Ds 2 to the point P 2 .

The direction of the arrows shows the direction of a

fall in the rate of interest.

conceived to be due to changes in investment brought
about by movements in the rate of interest.

Figure 10 represents the case discussed on p. 86, where
a fall in the rate of interest leads in the long period to a
fall in employment. Suppose that when equilibrium has
been reached, at P 15 to a certain rate of interest, a fall in
the rate of interest occurs. Employment will increase and
wages fall to a point, N l3 on the short-period demand curve
appropriate to P r As time goes by, if no further change
takes place employment will fall and wages rise to P 2 , the
point of long-period equilibrium corresponding to the
lower rate of interest. If the rate of interest is now raised
to its former level, employment will fall and wages rise to
a point, N 2 , on the short-period demand curve appropriate
to P 2 . As time goes by employment will increase and wages
fall until they are once more restored to the position P r

The short-period demand curve will normally be of the
form shown in Figure n, having a rapid change of slope
at the point, P, which corresponds to the level of employ-
ment, OE, at \^hich a preponderance of firms in the

130

ESSAYS IN THE JHEORY OF EMPLOYMENT

consumption industries ar^ working to their designed
capacity. If employment is increased beyond OE a sharp

FIG. 11

rise of prices and fall of real wages will occur. We have
suggested (p. 26) that OE represents in a certain sense
the optimum level of employment from a short-period
point of view.

The point of full employment represents the point
beyond which it is impossible for output to expand. But
an increase in output may be limited not only by the
available supply of labour but also by available plant and
equipment. It appears at first sight as though there must
be two limits the point of full employment and the point
of full capacity and that the second may be reached
before the first. This view, however, is not formally correct.
For at the point of full capacity the demand curve becomes
perfectly inelastic and must finally cut the x axis, while the
lower reaches of the supply curve must have a positive
elasticity and it must finally cut thejy axis. Thus if the two
curves have not intersected before the volume of employ-
ment is reached which corresponds to full capacity they
must intersect when it is reached. The point of full employ-
ment, therefore, cannot lie beyond the point of full capacity
and it is formally correct to say that the upper limit tc
the possible expansion of output is set by the point of full
employment.

DIAGRAMMATIC ILLUSTRATIONS 131

CHANGES IN UNEMPLOYMENT

In Figure 12 OE represents the actual amount of em-
ployment and OF, full employment. At first sight it seems

Fi<3. 12

natural to say that the amount of unemployment is
measured by EF. But this is a quantity which corresponds
to nothing in the experience of the community concerned,
for no one knows precisely how real wages would move if
employment were to expand. Let PW, parallel to the x
axis, cut the supply curve S, in W, and let the perpendicu-
lar WG cut the x axis in G. Then OG represents the
amount of work offered at the ruling real wage rate, PE
(or WG) . The amount of unemployment represented by
EG is the amount of labour, forthcoming at the ruling wage
rate, which cannot find employment. This corresponds
to a more or less definite element in experience in the
actual situation. There are a number of individuals who
have not got jobs who are anxious to have them, or who are
working part time when they would prefer to work full
time. The quantity EG is intended to represent the
amount of work which they are anxious to do. Their
anxiety may be more or less acute, and, as we have seen
(p. 6), a theoretical difficulty is presented by the fact that
it may be very acute without being strong enough to break
down the resistaAce to accepting a reduced money wage.

132

ESSAYS IN THE THEORY OF EMPLOYMENT

In any actual situation the precise magnitude of EG
cannot be discovered, but it corresponds to a common-

FIG. 13

sense conception, and substantial changes in it are easy to
detect.

Changes in unemployment can occur in three quite
distinct ways. There will be a reduction in unemployment,
first, when there is an increase in employment, due to an
increase in effective demand. This would be represented
in Figure 12 by a movement of E towards F and of P
towards Q,. In the case illustrated unemployment would
fall off less fast than employment increased, because the
fall in real wages accompanying an increase in employment
would increase the available supply of labour (p. 8 5 note).
Second, there may be a rise in the supply curve of labour.
This will occur when classes of individuals, such as children
or married women, are prohibited from working (p. 50)
or when there is an imposed reduction in the length of
the working day (p. 45). In any given situation a change
in the conditions of supply ,of labour is likely to react upon
the level of effective demand, but we will suppose for the
sake of illustration that conditions are such that the actual
amount of employment remains constant.

In Figure 13, the supply curve is altered from S^ to 8%.
Employment is constant at OE, and rfeal wages at EP.

DIAGRAMMATIC ILLUSTRATIONS

133

Full employment is altered from OF l to OF 2 . Unemploy-
ment^ reduced by G 2 G r

O E

Third, there may be a change in the demand curve for
labour, such as is due to a change in the degree of monopoly
or, in an open system, to a change in money wages or in
the exchange rate (p. 20). Once more we will suppose
that the reaction upon effective demand is offset, so that
employment remains constant.

In Figures 14 and 15 the demand curve falls from D l to
D 2 . Employment is constant at OE. Real wages fall from
E?! to EP 2 . Full employment is altered from OF^ to OF 2 .
Unemployment is reduced in Figure 14 by G 2 G l and in-
creased in Figure 15 by G 1 G 2 .

It is clear from this analysis that the removal of unem-
ployment may take deleterious as well as beneficial forms
and that a situation in which unemployment has dis-
appeared is not in all circumstances to be nreferred to one
in which it exists.

PART III
THE FOREIGN EXCHANGES 1

THE exchange rate is determined from day to day by
supply and demand of home currency in terms of foreign
currency. Each transaction is two-sided, and sales are
equal* to purchases. Any change in the conditions of
demand or of supply reflects itself in a change in the
exchange rate, and at the ruling rate the balance of
payments balances from day to day, or from moment
to moment.

The constituents of the demand for foreign currencies in
terms of home currency (or the supply of home currency
coming on to the exchange market) may be divided into
four groups. Foreign currency may be required (i) in
order to pay for goods or services purchased from foreigners
(or to make gifts to them), that is, in order to discharge
obligations in respect to visible and invisible imports,

(2) in order to make loans or purchase securities abroad,

(3) for speculative purposes, that is to say, in order to take
advantage of an expected reversal in the future course
of the exchange rate, (4) in order to remove funds from a
country in which political, fiscal or business prospects
appear threatening to one in which they seem relatively
secure. A fifth group is represented by official exchange
dealings, but, since our object is to discover the influences
determining the exchange rate in the absence of official

1 The subject-matter of this and the following essay has been the battle-ground
of innumerable controversies, and I have preferred (apart from one or two specific
points) to make no references to other writers, rather than to weary the reader with
continual acknowledgments and disagreements. It will be obvious that my main
endeavour is to elaborate the hints thrown out by Mr. Keynes in his Treatise on Money,
Chap. 21.

134

THE FOREIGN EXCHANGES 135

interference, we shall assume that no official dealings take
place, .except when special reference is made to them. 1

Interest on foreign capital invested in the home country
is most conveniently treated, in -the traditional way, as an
invisible import, since it represents a payment for the
services of foreign funds borrowed in the past. Day to day
fluctuations in the balances of professional exchange
dealers may be included in the third group, that is, as
speculative transactions, though they db not represent
speculation in the popular sense, but are part of the routine
business of the exchange market.

If gold is treated as a commodity, 2 'and if exchange
dealings as an instrument of official policy are ruled out of
account, the first four groups cover the whole field. The
third and fourth groups shade into each other, and the
second shades into both, for exchange prospects and
considerations of security influence foreign lending, while
funds removed to a foreign country for security will nor-
mally be lent at interest, though they may be used to pur-
chase gold or to make a deposit at a bank. 3 Thus the

1 An exchange rate which is considered undesirably high can be corrected by direct
intervention in the exchange market. But while the monetary authorities can always
sell an unlimited amount of their own currency they can only buy as much as their
holdings of gold or foreign assets permit. For this reason it is impossible to hold the
exchange rate indefinitely above the level determined by a free market without
resorting to complete control of all dealings.

2 In some contexts it is convenient to treat gold movements in a different way (see
General Theory,^. 335, and Treatise, p. 329). In accordance with our present scheme of
analysis an increment to the stock of gold inside the country must be regarded as
part of home investment and a decrement as disinvestment. This has the advantage
of putting the gold-producing countries upon the same footing as the rest, and of
putting a reduction in the stock of gold upon the same footing as a reduction in the
stock of any other commodity. But no point of principle is involved. An increase in
imports accompanied by an equivalent export of gold taken from stocks may be
treated either as a decrease in the balance of trade (foreign investment) without any
change in home investment, or as home disinvestment, without any change in the
balance of trade. Exports of newly-mined gold are in either case treated in the same
way as exports of any other commodity.

3 It is sometimes supposed that an increase in the prospective earnings of capital
in one country will lead to an increased desire on the part of foreigners to lend to it,
for instance, that a boom on Wall Street * attracts money* from Europe. But this
cannot occur when home and foreign speculators take an equally optimistic view of
prospects, for if they do flieprice of securities will be driven up to such a point as to

136 ESSAYS IN THE THEORY OF EMPLOYMENT

motives which govern the demand for currency for foreign
lending are inextricably bound up with the motives .which
govern exchange speculation and the panic moyement of
funds. It is therefore most convenient to distinguish only
two major categories in the balance of payments pay-
ments in respect to imports and exports, which represent
the income account or balance of trade, and payments in
respect to lending and borrowing (covering the whole of
the last three grOups distinguished above) which represent
the capital account or balance of lending.

Now since the balance of payments always balances
it follows that, fol 4 any section of time, payments in respect
to the balance of trade must be equal and opposite to
payments in respect to the balance of lending. Thus when,
over any period, the inhabitants of a country have col-
lectively a surplus of receipts from exports over payments
for imports (or positive balance of trade) 1 they must, over
the same period, on balance be lending (in the broad
sense) to the inhabitants of foreign countries an exactly
equal sum. Similarly a surplus of imports (or , negative
balance of trade) must be matched by an equal amount of
borrowing. This is merely a truism, but it is a truism
which provides the most convenient starting point for the
theory of the exchanges.

The same truism can be reached by another route,
which it may be instructive to turn aside for a moment to
follow. A positive balance of trade is equivalent to invest-
ment, from the point of view of the home country, and it
has the same influence as investment upon the level of
effective demand in the home country. It represents a
certain volume of demand for current home output with-

compensate for the improvement in their prospective yield, and no movement of
foreign funds will actually take place. Only if the change in opinion leaves foreigners
more bullish than home speculators will it lead to an inflow of foreign funds.

1 The movement of goods and performance of services do not normally synchronise
with the payments which are made in respect to them (cf. Haberler, Ttieory of Inter-
national Trade, p. 18). This fact is of significance in certain contexts, but in order to
avoid a cumbrous degree of exactitude we shall speak in wh^t follows of, for instance,
an increase in exports relatively to imports as an increase in the balance of trade.

THE FOREIGN EXCHANGES 137

out representing a supply of goods coming on to the home
market (for the trade balance represents the home incomes
earned, by selling to foreigners minus that part of home
incomes which is expended upon foreigners) and so gives
rise to secondary employment. Thus the trade balance
is one of the influences which determine the level of
income, and consequently the level of saving, in the home
country. Since the saving of a community, over any period
of time, is equal to its investment for that period, saving is
equal to* home investment plus or minus the trade balance.
New borrowing at home is equal to home investment,
while lending is equal to saving. Therefore foreign lending
is equal to the trade balance.

The truism, in the nature of the case, can throw no light
upon the manner in which, the equality is brought about.
It can only tell us that if in fact there is a change in the
balance of trade there must be an equal change in the
balance of lending. In some circumstances, as we shall
find, the two are directly bound together, but in general,
though there are always cross-connections between them,
they vary in response to independent sets of influences.
If I take it into my head to buy a foreign security there is
no reason why some compatriot of mine at the same
moment should decide to curtail his purchases of imported
goods .

The volume of imports and exports is determined by
tastes, techniques and resources the world over, and by
costs and incomes at home and abroad, which in turn are
determined by the levels of money wages and of effective
demand. The balance of lending is determined (given
wealth and incomes) by relative rates of interest at home
and abroad, and by all those considerations which may be
lumped together under the heading of 'the state of confi-
dence. 5 A fall in the home rate of interest (or a rise in
rates abroad) or the growth of dismal expectations about
home affairs will increase the desire to lend abroad. It is
by such diversft influences that the balance of trade and

138 ESSAYS IN THE THEORY OF EMPLOYMENT

the desire to lend are determined, and equality between
them is preserved not by any natural tendency for the, two
to vary consonantly, but by the mechanism of , supply
and demand. The theory of the exchanges may be regarded
as the analysis of the manner in which movements of the
balance of trade and the balance of lending are equated
to each other.

A change in the desire to lend abroad will tend to alter
the exchange rate. The reaction upon the balance of trade
of an alteration in ^he exchange rate must be examined at
some length. Suppose that, after a certain exchange rate
has been in force for some time, the amount which the
inhabitants of the home country desire to lend abroad
increases. At the ruling exchange rate the demand for
foreign currency exceeds the supply and the exchange rate
consequently falls. This has the effect of making home-
produced goods appear cheaper to foreigners and so
increasing the volume of exports. If the physical volume
of exports increases their home price cannot fall, therefore
the value of exports in terms of home currency must
increase. But the effect on imports is more complicated.
Foreign goods are now dearer at home, and while the
physical volume of imports purchased out of a given income
will decline, total expenditure upon them may increase.
Thus a decline in the exchange rate will not necessarily
increase the balance of trade. If the value of imports
(reckoned in home currency) increases by more than the
value of exports, then a fall in the exchange rate will
reduce the balance of trade.

The argument may be treated in terms of four elasticities:
the foreign elasticity of demand for exports, and the home
elasticity of supply (which is influenced by the home
elasticity of demand for exportable goods), the foreign
elasticity of supply of imports and the home elasticity of
demand for imports (which is influenced by the home

THE FOREIGN EXCHANGES 139

elasticity of supply of rival commodities) . x For brevity we
may speak of the demand for imports as 'home demand/
the demand for exports as 'foreign demand/ and so forth.

The actual change in imports and in exports which will
come about as the result of a change in the exchange
rate will depend partly upon the reaction on the demand
for imports, and on the supply of exports, of a change in
the balance of trade itself. For instance, an increase in
the balance of trade leads to an increasein home incomes,
and consequently to an increase in expenditure upon im-
ported goods; an increase in exports, or in home manu-
factures rival to imports, may lead to am increased impor-
tation of raw materials, while increased expenditure upon
home-produced goods may raise the supply curve of
exports. But these effects influence the magnitude, not the
direction, of the change in the balance of trade consequent
upon a fall in the exchange rate, 2 for the secondary effects
follow from the change in home incomes due to the change
in the balance of trade, and if the balance of trade does not
alter the secondary effects cannot occur. It is therefore
legitimate to discuss the initial effect upon the balance of
trade in terms of the four elasticities, abstracting from the
change in home incomes.

Let us first consider the export side of the balance sheet.
As we have seen, a fall in the exchange rate leads to an
increase in the value of exports in terms of home cur-
rency. The extent of the increase depends upon the elas-

1 It assumed throughout this and the following essays that elasticities of supply
are positive and of demand negative. Discussions of the magnitude of elasticities
must be taken to refer to their numerical, not their algebraical value.

2 This is not perfectly accurate, for qualitative differences between different types
of goods and of expenditure from different types of income may introduce complica-
tions into the. simple analysis here set out. For instance, suppose that the initial
effect of a fall in the exchange rate is to increase the value of both exports and imports
to the same extent, while export goods require a higher proportion of imported raw
materials than the home goods whose output declines when expenditure upon imports
increases. Then the initial effect of a fall in the exchange rate is to leave the balance
of trade unaltered while the final effect is to reduce it. The increase in home incomes
due to the increase in exports is then less than the reduction due to the increase in
imports.

1 4 o ESSAYS IN THE THEORY OF EMPLOYMENT

ticity of foreign demand (which must be reckoned in terms
of foreign currency). The increase in the value of exports
will be smaller the smaller is the foreign elasticity of
demand (given the home elasticity of supply). In the limit,
if the foreign demand is perfectly inelastic there will be no
increase in the volume of exports and consequently no
increase in their value.

Next consider the influence of home elasticity of supply.
If home supply iff perfectly inelastic the volume of exports
does not alter, their foreign price is unchanged and the
value of exports increases in proportion to the fall in the
exchange rate. If home supply is perfectly elastic, the home
price is constant and the price to foreigners falls in propor-
tion to the fall in the exchange rate. If the elasticity of
home supply lies between zero and infinity the home price
of exports is raised by an increase in their volume, and
their price to the foreigner consequently falls less than in
proportion to the fall in the exchange rate.

If the foreign elasticity of demand is equal to unity, so
that expenditure is constant in terms of foreign currency,
the value of exports is independent of the home elasticity
of supply and increases in proportion to the fall in the
exchange rate. If the foreign demand has less than unit
elasticity, the increase in the value of exports will be greater
the smaller is the increase in their physical volume, that is,
the smaller is their elasticity of supply. Thus, when the
foreign demand has less than unit elasticity the maximum
possible rise in the value of exports is that which is brought
about when their elasticity of supply is zero. The value of
exports then increases in proportion to the fall in the ex-
change rate. So long as the foreign demand has less than
unit elasticity any increase in the physical volume of exports
means that their value increases less than in proportion
to the fall in the exchange rate. On the other hand, when
the foreign demand has an elasticity greater than unity,
an increase in the volume of exports leads to an increase in
the foreign expenditure upon them, a^d the value of

THE FOREIGN EXCHANGES 141

exports increases more than in proportion to the fall in
the exchange rate. The increase in the value of exports is
then greater the greater is the elasticity of home supply.
In short, a high elasticity of home supply tends to reduce
or to enhance the increase in the value of exports induced
by a fall in the exchange rate according as the foreign
elasticity of demand is less or greater than unity.

The minimum effect of a fall in the exchange rate upon
the value of exports is produced when tl} foreign demand
has zero ^elasticity. There is then no increase in exports.
The maximum effect is produced when a perfectly elastic
foreign demand is combined with a perfectly elastic home
supply. The increase in the value of exports is then
indefinitely great.

We must now consider the import side of the balance
sheet. The value of imports in terms of home currency
will increase or diminish according as the elasticity of
demand is less or greater than unity.

If the foreign supply is perfectly elastic, so that the foreign
price of imports is constant, then their home price will rise
in proportion to the fall in the exchange rate; while if the
foreign supply is less than perfectly elastic a curtailment of
output will cause a fall in the foreign price, so that the
home price rises by less than the fall in the exchange rate.
It can be seen, therefore, that when the home demand has
less than unit elasticity, the value of imports will rise by
more, and when it has greater than unit elasticity, will
fall by more, the greater is the foreign elasticity of supply.

A fall in the exchange rate produces the maximum
increase in the value of imports when home demand is
perfectly inelastic. In this case the physical volume of
imports is constant, their foreign price is unchanged, and
both their price and their value in home currency are
increased in proportion to the fall in the exchange rate.
The maximum decrease is produced when a perfectly
elastic home demand is combined with a perfectly elastic
foreign supply. Tn this case imports are reduced to zero.

142 ESSAYS IN THE THEORY OF EMPLOYMENT

We must now combine the two sides of our balance
sheet. The relations between the various factors in the
problem are complicated, 1 but some simple generalisations
can be made. So long as the home demand for imports
has more than unit elasticity, a fall in the exchange rate
must increase the balance of trade, for the value of imports
falls, while the value of exports is at worst constant. If the
home demand for imports has less than unit elasticity, the
balance of trade< will still increase if there is a sufficient
increase in the value of exports, but if the elasticity of
foreign demand for exports is not sufficient to compensate
for a low elasticity of home demand, then a fall in the
exchange rate will reduce the balance of trade.

Before preceding further, the relative magnitudes of
the values of imports and exports must be considered.
For instance, if the elasticities are such that a fall in the
exchange rate brings about an equal proportional increase
in the value both of imports and of exports, then if imports
were equal to exports in the first instance, so that the
balance of trade was zero, it will remain zero when the
exchange falls. If it was positive in the first instance, it
will increase, for an equal proportional increase in exports

1 The general relationships can be expressed mathematically. Let I be the quantity
of imports, E of exports, p the home price of imports, and q the home price of exports.
Let h and e^ be respectively the elasticities of home demand for imports and of foreign
demand for exports, t\ h and ^ the elasticities of home supply of exports and of foreign
supply of imports. Consider the effect of a small fall in the rate of exchange in the
proportion k. Let the home price of exports rise by 5q. Then the fall in the foreign
price of exports is:

q (i - k)(q + dq)kq - 8q, k being small.

5E /5g

We now have i? fc = / ,

E / q

SE / dq

and e/= _./*--?.

Li / q

In the same way, 77 = =- / k

I/ p

, dl /dp

and u _ - -/_>

The increase in the balance of trade is (ESq -f q8E) (Ify -f- />$!), which can
be reduced to:

THE FOREIGN EXCHANGES 143

and imports entails a larger absolute increase in exports
if exports exceeded imports in the first instance. If the
balance of trade was negative in the first instance it will
be reduced by the fall in the exchange. 1 When the balance
of trade is zero in the first instance, then if the elasticity
of foreign demand for exports is greater than unity, a fall
in exchange rate must increase the balance of trade,
for the value of exports is increased more than in propor-
tion to the fall in exchange rate, while^in the worst case,
where hjome demand is perfectly inelastic, the value of
imports is increased only in proportion to the fall in ex-
change rate. If the elasticity of foreign demand for exports
is less than unity, the balance of trade will still increase
provided that the elasticity of home demand is sufficient to
compensate for the low elasticity of foreign demand for
exports.

It is now clear that the balance of trade may increase
with a fall in the exchange rate even if the elasticities of
foreign and home demand are both less than unity. In
the simple case, where trade is balanced in the first in-
stance*, and the elasticities of home and foreign supply are
both infinite, the balance of trade will increase or diminish
according as the sum of the elasticities of home and of
foreign demand is greater or less than unity, that is,
according as the deficiency below unity of the one is more
or less than offset by the excess above zero of the other. 2

1 The effect of inequality between Eg and Ip (in the notation of the foregoing foot-
note) can be shown most simply in the case in which the elasticities of foreign and
home supply are both infinite.

When ?// and f\h are both equal to infinity, the increase in the balance of trade
becomes:

k{Eq e, + lp e h - Ip}

It follows that, for the balance of trade to increase with a fall in exchange rate, it is
a sufficient, though not a necessary, condition that the elasticity of foreign demand
should exceed the ratio of imports to exports.

2 Cf. A. P. Lerner, The Economics of Control, p. 377. In this case not only are i^ and
v}h both infinite, but also Eg is equal to I/>, so that the expression for the increase in
the balance of trade becomes:

k Eg{e f + e h - 1}
which is positive or negative according as e/ -f- * H is greater or less than unity.

i 4 4 ESSAYS IN THE THEORY OF EMPLOYMENT

The repercussions of a change in the balance of trade
upon the home demand for imports and supply of exports
must be brought into account when the direction of the
change has been discussed in terms of the four elasticities.
The final change in the balance of trade, in either direc-
tion, will be smaller, the greater are the change in demand
for imports and the change in supply of exports brought
about by the changes in home activity and expenditure
which are due to t;he initial change in the balance of trade.
Further, since foreigners are impoverished or enriched by a
decrease or increase in the balance of trade of the world
with the home cojuntry, there is an additional secondary
reaction upon the foreign demand for exports, which also
tends to mitigate the change in the home balance of trade.

If, at a given exchange rate, the balance of trade falls
short of the balance of lending the exchange depreciates.
Under favourable conditions this leads to a sufficient
increase in the balance of trade to prevent any further fall
in the exchange rate. The most favourable conditions
from this point of view are, as we have seen, those in which
there is perfectly elastic foreign demand and home supply
of exportable goods. These conditions prevail as between
countries on the gold standard. In the home country
gold is on sale at a fixed price, while transport costs are
very low and do not rise, except in extreme circumstances,
with an increase in the volume of the commodity handled.
Supply is therefore perfectly elastic under normal con-
ditions. In foreign countries demand is perfectly elastic at
a fixed price. Any tendency for the exchange rate to fall
will therefore lead to an indefinitely large export of gold.
Similarly home demand and foreign supply are perfectly
elastic, so that any rise in the exchange rate would lead
to an indefinitely large import of gold. It is for this reason
that movements in the exchange rate cannot occur (beyond
the limits set by transport costs) so long as the gold standard
is maintained.

Other items in the trade balance have, certain peculiar

THE FOREIGN EXCHANGES 145

features of their own. Interest on foreign capital which is
fixed t in terms of home currency represents an export of
which the value cannot increase in response to a fall in
the exchange rate 1 (though, if debtors are distressed, the
reduction in the burden upon them brought about by a
rise in the exchange value of their currency may have an
important effect in preventing default). For a country
in whose total exports this item is an important element,
the beneficial effect upon the balance o trade of a fall in
the exchange rate is pro tanto diminished. From the point
of view of a debtor country, interest payments fixed in
terms of the creditor's currency represent an import which
rises in value in proportion to a fall in the exchange rate, 2
and if such obligations are considerable (and default is not
contemplated) exchange depreciation may be extremely
dangerous to the balance of trade. Obligations fixed in
terms of the debtor's currency represent, from the creditor's
point of view, an export whose value rises in proportion
to the fall in the exchange rate. 3 They tend, therefore, to
make the reaction of depreciation upon the balance of
trade favourable. From the debtor's point of view they
represent an import which is unaffected by a fall in the
exchange rate, 4 and so far as their influence goes, the
reaction of depreciation upon the balance of trade is
neutral.

A country whose main exports are manufactured goods
in which it has no monopoly will normally enjoy a fairly
elastic foreign demand, combined, except in boom con-
ditions, with a highly elastic home supply. Its exports will
therefore respond favourably to a fall in the exchange rate.
On the other hand, if its imports consist mainly of food and
raw materials which cannot be produced at home, the

1 Obligations to the home country fixed in home currency may be regarded from
a formal point of view as an export for which the foreign demand is perfectly inelastic.

8 They may be regarded as an import for which the home demand is perfectly
inelastic.

3 They may be regarded as an export of which the home supply is perfectly inelastic.

4 They may be regarded as an import for which the elasticity of demand is unity.
K

146 ESSAYS IN THE THEORY OF EMPLOYMENT

demand for imports is probably inelastic, while if it does
not represent a predominant part of the world market
the foreign supply will probably be highly elastic. The
effect of depreciation upon imports is then unlikely to be
favourable, and the benefit to the balance of trade of an
increased value of exports may be cancelled out by an
increased value of imports. 1 Moreover, if the balance of
trade does tend to increase, the extent of the increase will
be limited by the increased importation of raw materials
which results from increased activity.

A country which is dependent upon the production of
commodities (especially raw materials) of which it provides
a predominant part of the world supply will normally find
the demand for its exports relatively inelastic, for it has no
rivals at whose expense its sales can be increased, and
it is faced with the demand for each commodity as such.
In this case an inelastic home supply will be a source of
strength. 2 Countries of this type normally import manu-
factured goods for which the demand is likely to be rela-
tively elastic, compared to the demand for foodstuffs.
In respect to imports, therefore, the effect of deprecia-
tion upon the balance of trade is unlikely to be adverse.

In any given situation, with given wages, there will be,
for any one country, a certain rate of exchange at which
its balance of trade is at a maximum. This may be called .
the optimum exchange rate. It is the 'optimum 5 rate in a
strictly limited sense, for a fall in the exchange rate is
likely to raise the price of imports relatively to exports,
thus reducing real income per unit of output in the home
country, 3 so that the rate which maximises the trade balance
is by no means necessarily the most desirable rate from

1 Great Britain in 1931 escaped from the dangers of this situation because her
depreciation was mainly vis-a-vis rival manufacturing nations, while parity was main-
tained with countries responsible for a high proportion of her sources of raw materials.

2 For instance, the remarkable steadiness of Australia's output of wool was an
important factor in the benefit which she derived from depreciation in 1 93 1 . In default
of a naturally inelastic supply monopolistic restriction schemes are widely resorted
to by raw-material-producing countries.

8 See below, p. 164.

THE FOREIGN EXCHANGES 147

every point of view. Moreover, a change in the exchange
rate i of one of the major countries produces so many
reactions upon the rest of the world, and such far-reaching
economic and political effects, that it would be absurd to
treat it merely in terms of elasticities of supply and demand.
But such treatment is a necessary part of the more general
discussion of exchange problems, and it is to this narrow
sphere that the present analysis is confined.

If the exchange rate stands at the optimum level, any
chance fell will precipitate a progressive decline, for each
fall in the rate reduces the trade balance and promotes
a further fall. In the absence of control* the exchange rate
is stable only so long as it stands above the optimum level.
But the value of the optimum rate largely depends upon
the length of the run which is feeing considered. From the
point of view of very short-period reactions to a fall in the
exchange rate, both the foreign elasticity of demand for
exports and the home elasticity of demand for imports are
likely to be very low (apart from gold), even when over
a longer run they would prove to be great, for the fall in
the price of the one and rise in the price of the other takes
time to produce its effect upon the decisions of purchasers,
while, if prices are agreed in terms of the exporter's cur-
rency, the force of inertia (and prearranged contracts)
delays the rise in the value of exports. 1 Thus it appears
at first sight that from the point of view of a very short run
the exchange rate can never be above the optimum, and
that any country which has abandoned the gold standard
must be in chronic danger, no matter how strong its long-
period position, that the smallest increase in the balance
of lending will precipitate a sudden collapse in the exchange
rate.

1 If prices are agreed in terms of the importer's currency the short-period reaction
of a fall in the exchange rate is favourable. For import prices fail to rise, so that in-
elastic home demand is innocuous, while, in the first instance, the value of exports
rises in proportion to the fall in the exchange rate. When the Belga was devalued
exporters were urged to continue to charge the same foreign prices and not to increase
their output. The effect upon the balance of trade is the same, in such a case, as
though home supply weretperfectly inelastic.

148 ESSAYS IN THE THEORY OF EMPLOYMENT

But against this danger there are two important safe-
guards. So long as any country in the world adheres to
the gold standard there is one commodity for which even
the short-run demand is perfectly elastic, while if thete is a
market in gold in the country whose exchange is falling
the supply of this export will be highly elastic, though less
elastic than when it is officially on offer at a fixed price. 1
Exports of gold will thus serve as a stop-gap, and prevent
the exchange rate from collapsing at a breath.

Further, a fall in the exchange rate which is not expected
to last will call professional speculators into action. Pur-
chases of the depreciated currency, representing a form of
foreign borrowing, will bridge the gap in the balance of
payments and prevent the exchange rate from falling
beyond the level at which it is expected later to come to
rest. Thus time will be allowed for a moderate fall in the
exchange rate to produce its effect upon the balance of
trade, and a slight fall will not necessarily lead to an
immediate collapse.

On the other hand, as is only too well known, if specula-
tors read a slight fall as a sign that a further fall is to be
expected, a violent increase in foreign lending (in the wide
sense) will take place and the balance of trade will have no
time to react to an initial fall in the exchange rate before
a further fall takes place. In this case only official interven-
tion can prevent a sudden collapse.

What happens if there is no intervention, while foreign
lending remains constant or increases as the exchange
rate falls below the level of the short-period optimum?
The rate is then sent hurtling towards zero. But on its
way thither it must necessarily pass through a pessimum
point (at which the balance of trade is a minimum) and
come to rest somewhere below it. For a sufficiently
violent rise in the price of imports must ultimately choke
demand, and even if exports fail to react, in the flurry of

1 See Einzig, Bankers , Statesmen and Economist ,*p. 86.

THE FOREIGN EXCHANGES 149

the moment, the value of imports must somewhere begin
to fell off.

A change in the desire to lend abroad can bring about a
change in the balance of trade (and consequently in the
actual rate of lending) only by way of its effect in altering
the exchange rate. But a change in the balance of trade
produces a direct effect upon the balance of lending.
The rate of saving in the home country exceeds or falls
short of the rate of home investment according as the
balance of trade is positive or negative. In the normal way
a part of the increase in the wealth of individuals in the
home country represented by home saving will be used to
acquire foreign securities or to* make loans abroad. Thus,
when the balance of trade increases, and home saving con-
sequently increases, this in itself will lead to an increase in
foreign lending. Similarly, when the balance of trade
falls off, lending by the world to the home country is
directly increased. To look at the same thing in another
way, if the balance of trade falls off, there is an increase in
the excess of the rate at which home securities (represent-
ing borrowing for home investment) are coming on to the
market over the rate at which wealth at home is accumu-
lating, while, at the same time, in the rest of the world
there is an increase in the excess of the rate at which wealth
is accumulating over the flow of new securities, and the
world is inclined to buy home securities at a greater rate
than before.

If the world capital market were perfect, so that owners
of wealth, the world over, were completely indifferent as
between home and foreign securities, then, when the home
balance of trade falls off, the excess new savings of tlie world
would be devoted to buying the excess of new home
securities over new home savings, without any change in
relative interest rates, and the rate of exchange would remain
in equilibrium 3r spite of the fall in the balance of trade.

1 5 o ESSAYS IN THE THEORY OF EMPLOYMENT

This is normally the case as between different parts of the
same country.

But the international capital market is not perfect, and,
if foreigners are to be attracted to buy home securities at
an increased rate, the home rate of interest must rise
relatively to the foreign rate. If the home rate of interest
does not rise sufficiently, foreign lending to the home
country will fail to increase by as much as the balance of
trade has fallen, ap,d the rate of exchange will consequently
fall.

4

We must now consider the effect of a change in the home
rate of interest upon the rate of exchange. Suppose that a
rise in the home rate of interest occurs, other things remain-
ing the same. A rise in the home rate of interest produces
its effect upon the exchange in three stages.

The first stage is represented by the additional foreign
borrowing (or reduction of lending) which is produced by
a rise in the relative rate of interest at home. The incre-
ment of borrowing may be divided into two parts, a small
increase in the share of world savings devoted to the home
country, which will persist (apart from unfavourable
reactions upon confidence) so long as the rise in the relative
home rate is maintained, and a larger, non-recurrent
movement due to the transfer of funds, formerly held
abroad, to the home country. Each transfer can only
affect the exchange rate as it is made, and to maintain a
given volume of transfers over an appreciable length of
time the relative rate of interest would have to rise con-
tinuously. Thus the balance of payments is not in equilib-
rium to a given exchange rate and interest rate unless no
transfers are taking place. 1

1 Mr. Sayers ('Japan's Balance of Trade,' Economica, February 1935, p. 52) suggests
that the exchange is in equilibrium when no short-term lending or borrowing is
taking place. For practical purposes short-term borrowing provides a useful index
of the purely transfer element in the international movement of funds, though the
two are not completely identical.

Equilibrium, in the sense that no transfer borrowing or gold movements are taking

THE FOREIGN EXCHANGES 151

There is here a close analogy to gold movements, which
also constitute a symptom of disequilibrium in the balance
of payments. If the home authorities are subject to the
legal obligations of the gold standard or, under influence
of more general considerations, desire to maintain a given
exchange rate it is the objective of their policy to establish
equilibrium in the balance of payments at the exchange
rate which they desire to establish, that is to say, to
create a situation in which neither gold movements nor
transfer lending are taking place at the desired exchange
rate.

The second stage in the operation ofihe rate of interest
is its effect upon the balance of trade. A rise in the home
rate of interest will curtail investment and so lead to a
decline in activity and incomes in the home country.
Expenditure upon imports will therefore fall off. Foreign
export industries will contract, and the consequent
decrease of incomes and expenditure in the rest of the world
will reduce the demand for goods exported by the home
country. But a part of the reduction in foreign incomes
will be subtracted from saving, and even if the supply
price of exports in the home country is unaffected it is
impossible that exports should be curtailed to the same
extent as imports. Moreover, the supply price of a given
volume of exports is likely to be reduced, for exportable
goods are partly consumed at home and a reduction in
home demand will increase the supply available for export,

place, is not the same thing as full equilibrium. Full long-period equilibrium of the
balance of payments does not obtain so long as any lending or borrowing is taking
place at all. For so long as borrowing is taking place the invisible imports represented
by interest payments are mounting up, and as time goes by persistent borrowing will
lead, other things equal, to a gradually falling balance of trade. (If, as sometimes
occurs, the borrowing corresponds to home investment which would not have taken
place without it, other things are not equal, for increased capital equipment will have
its reaction upon the other items in the balance of trade.) Full long-period equilibrium
is reached only when investment and saving are equal to zero, and imports are equal
to exports a state of affairs which has never been attained in actuality. The present
discussion is not concerned with these remote effects, and must be regarded as applying
to a length of run within which the accumulation of interest payments is small relatively
to current borrowing.

152 ESSAYS IN THE THEORY OF EMPLOYMENT

while services common to all industries, such as transport,
are likely to become cheaper to the exporters when the
total demand for them is reduced. Thus, in spite of the,
decrease in foreign demand, the volume of exports may
actually increase. In short, a decline of effective demand
at home tends to decrease imports relatively to exports
and so to increase the balance of trade.

Equilibrium with the given exchange rate is reached
when, at a constant rate of interest, the balance of trade,
excluding gold, is brought to equality with the balance of
recurrent lending, and no transfer of funds or movement
of gold is taking place. But the position is only attained at
the expense of unemployment at home, and any reduction
in the rate of interest, by stimulating activity, would set
up a tendency for the exchange to fall. The third stage is
not reached until increased unemployment has brought
about a fall in money wages in the home country.

The effect of an all-round reduction in money wages in
the home country upon the balance of trade is precisely
similar, apart from obligations which are fixed in terms of
home currency, to the effect of a corresponding fall in
the exchange rate, for both represent a decline in home
incomes and prices measured in terms of foreign currencies.
The effect upon the value of imports and of exports in
terms of foreign currency is the same for a fall in home
wages as for a fall in the exchange rate, while the home
purchasing power of a given amount of foreign currency
increases equally in each case. Obligations fixed in terms
of home currency introduce a difference between the two,
for while these are unaffected by a fall in the exchange rate,
the real burden of payments, and the real value of receipts,
are increased by a fall in home wages. 1 There is a further
difference between the effect of pressure upon the exchange

1 This applies equally to internal and external obligations, and the increase in
the share of rentiers in the national income, brought about by a fall in money wages,
may have some reaction upon the demand fpr imports, which would introduce a
further difference between the effect on the balande of trade of a fall in monev watres
and of a corresponding fall in the exchange rate.

THE FOREIGN EXCHANGES 153

rate and of pressure upon the level of money wages which
is of the utmost practical importance. While a fall in the
Qxchange will have an automatic and equal effect on the
relationship of all home prices to foreign prices, a fall in
money wages is never spread evenly over all industries
and relative prices inside the home country are never
unaffected by it. But for the purposes of our present
formal treatment we will consider a case in which the
unemployment caused by a rise in the rat of interest brings
about anequal proportionate fall in all wages rates.

Just as, with given money wage rates, there is an opti-
mum exchange rate, at which the balance of trade is at a
maximum, so, with a given exchange rate, there is an
optimum level of money wages. In circumstances in which
a fall in the exchange rate would lead to an increase (in
terms of home currency) in the "value of imports greater
than the increase in exports (apart from monetary obliga-
tions 1 ), an equivalent fall in wages would lead to a decline
in the value of exports greater than the decline in value of
imports. In such circumstances it is a rise, not a fall, in
the level of wages which would redress the balance of
payments. 2 But we may suppose for our present purpose
that the pre-existing level of money wages was above the
optimum in this sense. A fall in money wages will then
increase the balance of trade corresponding to a given
level of effective demand at home. If the interest rate is
maintained at its higher level after wages have fallen the
exchange will tend to appreciate; the rate of interest may

1 A fall in the exchange rate will have a more favourable effect upon the balance
of trade (reckoned in terms of home wage units) than a corresponding fall in wages
where payments to foreigners fixed in terms of home currency are an appreciable
element in imports, and a less favourable effect when receipts paid in terms of home
currency are an appreciable element in exports.

2 If the exchange has once been allowed to fall below its long-run optimum level,
the authorities are landed in an extremely awkward situation. For, while a rise in the
rate of interest will produce a beneficial effect upon the exchange at the first stage of
its operation and, by reducing employment, at the second stage, its effect at the third
stage will make matters worse than ever. In such a case curtailment of imports (by
tariffs and so forth) and of foreign lending, combined with direct intervention in the
exchange market, will fJrqyide the only remedy.

i 5 4 ESSAYS IN THE THEORY OF EMPLOYMENT

then be lowered and a recovery of employment allowed to
take place.

It is in this way that a tendency for the exchange rate to
alter can be offset by appropriate changes in the home
rate of interest.

It is now obvious that there is no one rate of exchange
which is the equilibrium rate corresponding to a given
state of world demands and techniques. In any given
situation there is an equilibrium rate corresponding to each
rate of interest and level of effective demand, and any rate
of exchange, within very wide limits, can be turned into
the equilibrium rate by altering the rate of interest appro-
priately. Moreover, any rate of exchange can be made
compatible with any rate of interest provided that money
wages can be sufficiently altered. The notion of the equilib-
rium exchange rate is a chimera. The rate of exchange,
the rate of interest, the level of effective demand and the
level of money wages react upon each other like the balls in
Marshall's bowl, and no one is determined unless all the
rest are given. 1

It will be observed that in the foregoing argument the
operations of the gold standard are treated in the same
terms as the workings of so-called free exchanges. The
only difference between the two is that under the gold
standard the authorities are committed to one particular
exchange rate so that the equilibrium of the balance of
payments must be preserved in face of changing conditions
entirely by inducing changes in the level of incomes, and
not at all by allowing variations in the exchange rate, while
under free exchanges the authorities have some measure
of latitude in their choice between the two methods of
adjustment.

For a country in which money wages do not readily

1 One more ball in the bowl is represented by expectations as to the future course
of the exchange rate; see p. 148,

THE FOREIGN EXCHANGES 155

yield to the pressure of unemployment the gold standard
can be maintained, in an era of rapid change, only by
Hjeans of recurrent periods of severe unemployment, 1 and
it is the Realisation of this fact which has in recent years so
much impaired the popularity of the gold standard.

1 The monetary history of Great Britain between 1925 and 1931 is the history of a
struggle between the level of money wages and the rate of exchange. It was appro-
priate that the final collapse of the gold standard should have been brought about
by a protest against cuts in pay.

BEGGAR-MY-NEIGHBOUR REMEDIES* FOR
UNEMPLOYMENT

FOR any one country an increase in the balance of trade
is equivalent to an increase in investment and normally
leads (given the level of home investment) to an increase
in employment. 1 An expansion of export industries, or of
home industries rival to imports, causes a primary increase
in employment, while the expenditure of additional
incomes earned in these industries leads, in so far as it
falls upon home-produced goods, to a secondary increase
in employment. But an increase in employment brought
about in this way is of a totally different nature from an
increase due to home investment. For an increase in home
investment brings about a net increase in employment
for the world as a whole, while an increase in the balance
of trade of one country at best leaves the level of employ-
ment for the world as a whole unaffected. 2 A decline in
the imports of one country is a decline in the exports of
other countries, and the balance of trade for the world as a
whole is always equal to zero. 3

In times of general unemployment a game of beggar-
my-neighbour is played between the nations, each one
endeavouring to throw a larger share of the burden upon
the others. As soon as one succeeds in increasing its
trade balance at the expense of the rest, others retaliate,
and the total volume of international trade sinks continu-
ously, relatively to the total volume of world activity.
Political, strategic and sentimental considerations add fuel

1 See below, p. 1 59, note, for an exceptional case.

2 Unless it happens that the Multiplier is higher than the average for the world
in the country whose balance increases.

8 The visible balances of all countries normally add up to a negative figure, since
exports are reckoned f.o.b. and imports c.i.f. But this is compensated by a correspond-
ing item in the invisible account, representing shipping and Dandling costs.

156

BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOYMENT 157

to the fire, and the flames of economic nationalism blaze
ever higher and higher.

Jn the process not only is the efficiency of world produc-
tion impaired by the sacrifice of international division of
labour, but the total of world activity is also likely to be
reduced. For while an increase in the balance of trade of
one country creates a situation in which its home rate of
interest tends to fall, the corresponding reduction in the
balances of the rest tends to raise their rates of interest, and
owing to the apprehensive and cautious tradition which
dominates the policy of monetary authorities, they are
chronically more inclined to foster a rise in the rate of
interest when the balance of trade is reduced than to permit
a fall when it is increased. The beggar-my-neighbour game
is therefore likely to be accompanied by a rise in the rate
of interest for the world as a whole and consequently by a
decline in world activity.

The principal devices by which the balance of trade
can be increased are (i) exchange depreciation, (2) reduc-
tions in wages (which may take the form of increasing
hours of work at the same weekly wage), (3) subsidies to
exports and (4) restriction of imports by means of tariffs
and quotas. To borrow a trope from Mr. D. H. Robert-
son, there are four suits in the pack, and a trick can be
taken by playing a higher card out of any suit.

Before proceeding any further it is necessary to make a
digression, for it has sometimes been denied that the restric-
tion of imports will increase home employment. 1 This
view appears to arise from a confusion as to the nature of
the classical argument for free trade. The classical argu-
ment states that (with certain well-known exceptions)
the pursuit of profit will bring about the specialisation of
resources and the distribution of trade between nations in
such a way that the maximum of efficiency is achieved.

1 Sec General Theory, p. 334. Mr. Keynes offers himself as a sacrifice. But (pace Sir
William Beveridge) it was never the orthodox view that a tariff cannot lead to an
increase in employment hMfre short period; see Pigou, Public Finance^ p. 224.

158 ESSAYS IN THE THEORY OF EMPLOYMENT

Any arbitrary interference with the channels of trade
will therefore lead to a decline in efficiency, and a reduc-
tion in the amount of output obtained from a given amount
of resources. This argument, on its own ground, is unexcep-
tionable. But in the nature of the case it can throw no
light upon the division of a given total of employment
between nations. It tells us that, with given employment,
output per head will be higher when trade is free. It
cannot tell us th?t when one country increases its share in
world employment, at the expense of reducing output per
unit of employment, its total output will be reduced.
Still less can it tell us that employment in any one country
cannot be increased by increasing its balance of trade.
Indeed it is obvious to common sense that a tax upon im-
ported goods will lead to an increase in the output of rival
home-produced goods, just as a tax upon any commodity
will stimulate the output of substitutes for it.

The popular view that free trade is all very well so long
as all nations are free-traders, but that when other nations
erect tariffs we must erect tariffs too, is countered by the
argument that it would be just as sensible to drop rocks
into our harbours because other nations have rocky coasts. 2
This argument, once more, is unexceptionable on its own
ground. The tariffs of foreign nations (except in so far as
they can be modified by bargaining) are simply a fact of

1 The argument is backed up by the contention that 'exports pay for imports,'
see, e.g., Beveridge and others, Tariffs: the Case Examined, chap. vi. It is admitted that
in some circumstances imports may be curtailed without exports falling to an equal
extent, but this entails an increase in foreign lending, and it is argued that if foreign
lending increases, home investment must decline (loc. '/., p. 57). Now when the
imposition of a tariff increases the balance of trade the increase in foreign lending
which is required to prevent a rise in the exchange rate is brought about by a fall in
the home rate of interest, and this is calculated to increase, not diminish, the volume
of home investment. The flaw in the argument consists in overlooking the fact that
an increase in home income will increase saving, so that increased foreign lending is
not made at the expense of lending at home.

The classical, as opposed to the neo-classical, argument is usually set out upon the
assumption that full employment is the normal state, and in the classical system of
analysis the question of a beggar-my-neighbour increase in home employment does not
arise.

2 Beveridge, op. '/., p. no.

BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOYMENT 159

nature from the point of view of the home authorities,
and the maximum of specialisation that is possible in face
3 them still yields the maximum of efficiency. But when
the gan\e of beggar-my-neighbour has been played for
one or two rounds, and foreign nations have stimulated
their exports and cut down their imports by every device
in their power, the burden of unemployment upon any
country which refuses to join in the game will become
intolerable and the demand for some f9rm of retaliation
irresistible. The popular view that tariffs must be
answered by tariffs has therefore much practical force,
though the question still remains open from which suit
in any given circumstances it is wisest to play a card.

Exchange depreciation and a reduction in the level of
money wages lead to an increase in the balance of trade,
in the manner which has already been discussed, 1 provided
that each stands above the optimum level. 2 A subsidy to
exports will increase the balance of trade provided that
foreign demand has an elasticity greater than unity, 3 while
restriction of imports by quotas will increase the balance
of tradfe provided that home demand has an elasticity
greater than unity. These four expedients are thus all
limited in their scope. A tariff reduces the volume of
imports, and tends to reduce their foreign price, even when
home demand is inelastic. Total expenditure by home
consumers upon imports, including tax payments, may
increase, but the payment to foreigners must be reduced.
Tariffs thus provide an expedient for increasing the balance
of trade which can still be used when all else fails.

We must now consider the effect upon home employ-
ment of an increase in the balance of trade brought about

1 See page 138. 2 See p. 146.

8 When the foreign demand is inelastic a tax on exports (as in Germany in 1922)
or restriction of output (as in many raw-material-producing countries in recent
years) will increase the balance of trade (cf. p. 146), while at the same time reducing
the amount of employment in the export industries, and increasing the ratio of profits
to wages in them. In these circumstances, therefore, an induced increase in the balance
of trade may be accomoanied bv no increase, or even a decrease, in the level of
employment.

i6o ESSAYS IN THE THEORY OF EMPLOYMENT

by each of the four expedients. To simplify the discussion
we may postulate that the funds necessary for a subsidy are
raised, or the receipts from import duties expended, in
such a way as not to interfere with the distribution of
income or to alter thriftiness in the home country. 1 Each
expedient must be supposed to produce its own full effect.
For instance, it must not be supposed that the influence of
a fall in the exchange rate on the balance of trade is
counteracted by *a rise in money wages, or that a tariff
leads to a rise in the exchange rate.

A fall in the exchange rate, or in money wages, causes
a primary increase in employment in export industries,
and in industries producing goods rival to imports. For
a given increase in the value of exports (in terms of home
wage units) the increase in employment will be greater
the greater is the elasticity of supply, and for a given
decrease in the value of imports it will be greater the greater
is the elasticity of foreign supply and the greater is the
elasticity of supply in the rival home industries. 3 It is
possible that an increase in the balance of trade may lead
to no primary increase in employment. .For instance,
suppose that the elasticity of home supply of export goods
is zero and the elasticity of demand for import goods
unity. Then a fall in the exchange rate will lead to a
proportional increase in the value of exports, without any
increase in their volume, and consequently without any
increase in employment in the industries producing them,

1 The manner in which funds are raised or receipts expended is, of course, of the
utmost importance, but analysis of the effects of changes in fiscal policy on employ-
ment can easily be superimposed upon the analysis here set out. For instance, if
receipts from import duties are paid into a sinking fund, or used to relieve taxation
on the rich in such a way as to increase their savings, there will be an increase in thrifti-
ness which will counteract the effect upon employment of increased foreign invest-
ment.

8 If the elasticity of demand for imports is less than unity, there will be a primary
decrease in employment in these industries, since additional expenditure upon imports
will be made at their expense, but in this case a given increase in the balance of trade
must entail so much the greater increase in exports.

* This generalisation can be made applicable to the exports and imports repre-
sented by foreign obligations if the elasticities concerned are treated in the manner
suggested in the footnotes to p. 145.

BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOYMENT 161

while the value of imports and the output of rival com-
modities will be unchanged.

In the case of a subsidy the primary increase in employ-
ment is in the export industries alone, 1 while in the case of
a tariff the primary increase is in the industries rival to
imports 2 and in the industries benefited by the expenditure
of the receipts from duties. 3 In the case of quotas the
primary increase is in the rival industries alone.

In each case, the increase in incomes dije to the increased
balance of trade will lead to secondary employment.
Thus even when there is no primary increase in employ-
ment at all, total employment will increase as a result of
the increased balance of trade. The lower are the elastici-
ties of supply in the industries primarily affected the greater
will be the increase in profits, relatively to wages, in them,
and the smaller the increase in expenditure coming from
them. Thus the secondary increase in employment is
likely to be smaller the smaller is the increase in primary
employment.

We must next consider the effect of the various
expedients upon real income per unit of employment.
Output per unit of employment normally falls off as em-
ployment increases. For a given increase in employment
the decline in output per unit of employment will be greater
in the case of subsidies, tariffs or quotas than in the case of
exchange depreciation or a fall in wages, since advance is
being made upon a narrower front. This is merely another
way of stating the classical argument that the mal-distribu-
tion of resources due to an artificial stimulus of particular
industries leads to a decline in output for a given level of
employment.

The change in income per unit of employment will also

1 While there may be a primary decrease in employment in industries whose
costs are raised as a result of the increase in output of export goods or whose receipts
are reduced by the collection of funds for the subsidy.

a While there may be a primary decrease in employment in the industries whose
costs are raised.

8 In general, the more elastic is the demand for imports the larger will be the
increase in the output of Jjie rival industries and the* smaller the proceeds of the duties.
Cf. above, p. 160, note, s

L

162 ESSAYS IN THE THEORY OF EMPLOYMENT

be influenced by the effect of the various expedients upon
the terms of trade. An improvement in the terms of trade,
that is, a rise in the price of exports relatively to the prijgg
of imports represents an increase in incomes, pef* unit of
employment, earned in export industries, relatively to the
cost of imported commodities. If the total value of imports
and of exports is more or less commensurate an improve-
ment in the terms of trade will therefore bring about a
rise in the averag real income per unit of employment for
the country as a whole.

A fall in money wages, which affects all industries
equally, is equivajent, as we have seen, to an equal pro-
portional fall in the .exchange except in respect to obliga-
tions fixed in terms of home currency. 1 Abstracting from
them for the moment, we may conduct our discussion in
terms of exchange depreciation alone, the argument being
made applicable to a fall in wages by means of reckoning
prices and incomes in terms of home wage units.

A fall in the exchange rate, which stimulates the output
of export goods and reduces the demand for import goods,
leads to a fall in the world price of both types of goods,
and a rise in the home price. Since the prices of both types
of goods move in the same direction it is impossible to say
out of hand what the effect will be upon the terms of trade.

The fall in the world price of export goods in the first
instance will be greater the less elastic is the foreign demand
for them, and the more elastic is the home supply; while
the fall in the price of import goods will be greater the
more elastic is the home demand and the less elastic is
the foreign supply. It can be seen that if the elasticity
of foreign demand for exports is equal to the elasticity of
foreign supply of imports, while the elasticity of home
supply of exports is equal to the elasticity of home demand
for imports, the initial effect of a fall in the exchange rate
will be to move both sets of prices to the same extent,
so that the terms of trade are unchanged. Further, if

5 P. 152.

BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOYMENT 163

the foreign elasticity of supply exceeds the foreign elasticity
of derjiand in the same proportion as the home elasticity
eC^demand exceeds the home elasticity of supply, the terms
of trade are unchanged, 1

In general, each country is more specialised in respect
to the goods which it produces than in respect to the goods
which it consumes, so that any one country plays a more
dominant role in the world supply of those goods which it
exports than it plays in the world market for those goods
which it ifnports. In general, therefore, the world demand
for the exports of one country is less elastic than the world
supply to it of those goods which it imports. So far as
the foreign elasticities are concerned, there is thus a strong
presumption that a fall in the exchange rate will turn the
terms of trade in the unfavourable direction.

Each country imports a large number of commodities
which cannot be produced at home, so that the elasticity
of demand for imports tends to be low. The elasticity of
supply of exports will depend upon the particular types of
goods in question, and upon the general state of trade.
In slump conditions, such as prevail when the game of
beggar-my-neighbour is most in vogue, the elasticity of
supply of all commodities, except certain agricultural
products, is likely to be high. It is thus only in exceptional
cases that the home elasticity of demand can exceed the
home elasticity of supply to a sufficient extent to com-
pensate for the excess of the foreign elasticity of supply over
the foreign elasticity of demand, and in general a fall in
the exchange rate must be expected to cause a deterioration
in the terms of trade.

An exceptional case would occur if the home supply of
exportable goods were perfectly inelastic. There would
then be no fall in the world price of exports, while unless

1 Using the notation of p. 142, note, the adverse change in the terms of trade is

oto So i f fif c/ \

- --- * which is equal to k { --- ). Thus the change in the terms of trade

P q M V/1+*?/ *to+*//

is adverse or favourable afcording as ~ is greater or less than ~A

164 ESSAYS IN THE THEORY OF EMPLOYMENT

either home demand for import goods is perfectly inelastic
or the foreign supply of them perfectly elastic, there will
be some fall in the price of imports, and the terms of trade
will become more favourable when the exchange rate
falls. Thus, as we have already seen, 1 for an agricultural
country which produces a considerable proportion of the
world supply of some commodity, the drawbacks of an
inelastic world demand for its exports may be overcome
by a sufficiently inelastic home supply. A country for which
an inelastic foreign demand is combined with' a highly
elastic home supply will suffer a serious deterioration in
the terms of tracte as a result of exchange depreciation.

The importance of the home country in world markets
will also affect the result. The change in world prices
brought about by exchange depreciation will in general
be smaller the smaller is the country concerned, and the
narrower will be the range of the possible changes in the
terms of trade. A large country is likely to suffer a greater
deterioration in the terms of trade, when its exchange
depreciates, than a small country, but at the sapie time
it is only for a very large country that a favourable move-
ment in the terms of trade can possibly occur, for it is
only a large country which can exercise an appreciable
influence on the world prices of the goods which it imports.

The effect upon the terms of trade of a fall in money
wages differs from the effect of depreciation in so far as
there are foreign obligations fixed in terms of home
currency. These are unaffected by a fall in the exchange
rate, while a fall in wages raises the cost of payments and
the value of receipts in terms of home wage units. Thus, in
so far as payments fixed in terms of home currency are an
appreciable element in invisible imports, the deleterious
effect of a fall in wages upon the terms of trade will be
greater than the effect of a corresponding depreciation
in the exchange, while a given increase in the balance
of trade, in terms of wage units, will reauire a larger fall

1 P. 146.

BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOYMENT 165

in wages, and so entail larger changes in the 'prices of other
imports and exports. In so far as receipts fixed in terms of
home* currency are an appreciable element in invisible
exports, *the deleterious effect of a fall in the exchange rate
will be greater.

A subsidy to exports leads to a fall in the world price of
export goods which will be greater the less elastic is foreign
demand and the more elastic is home supply. In so far
as the price of import goods is affected at all, it must be
raised. TJbe output of export goods is increased, and their
price in the home market, in which they are not subsidised,
is raised, 1 so that the price of imports ^hich are rival in
the home market to exportable goods may be raised. A
subsidy to exports therefore causes an unfavourable
movement in the terms of trade. 2 In this respect a subsidy
is necessarily more deleterious thkn exchange depreciation
or a fall in money wages.

A tariff leads to a fall in the world price of import goods,
which will be greater the less elastic is foreign supply and
the more elastic is home demand. In so far as it affects
the price of exports it must raise them. Raw materials
entering into export goods may be subject to duties, while
the increase in the output of home goods which are sub-
stitutes for imports may raise the price of the exportable
goods. A tariff therefore has a favourable effect upon the
terms of trade.

Neither a tariff nor a subsidy can normally be applied
to the invisible exports and imports (with the exception
of shipping services). Where it is possible to increase the
invisible balance by means of exchange depreciation
without any adverse effect upon the terms of trade (for

1 Services such as transport must be regarded as exports in so far as they enter
into the production of export goods.

2 Income per unit of output in the export trades is not reduced, but real income
per unit of output for the country as a whole is reduced by the levy of funds to pay
the subsidy.

3 This is known as 'making the foreigner pay the tax.* If foreign supply is perfectly
inelastic, price to the home consumer is not raised by the import duty at all and *the
foreigner pays the whole *f the tax.*

1 66 ESSAYS IN THE THEORY OF EMPLOYMENT

instance when the main invisible export consists of receipts
fixed in terms of foreign currency), the advantage of a
tariff, as opposed to exchange depreciation, is pro tanto
diminished, and the disadvantage of subsidies increased.

The restriction of imports by means of quotas does not
have the same effect upon the terms of trade as a tariff,
since it leads to a rise in the home price of import goods,
while preventing the restriction in home consumption
from lowering the foreign price. A quota upon imports
has much the same effect as an increase in the degree of
monopoly amongst foreign suppliers. It leads to a
deterioration in the terms of trade, while the benefit from
the raised price to the home consumer, which goes to the
exchequer under a tariff, goes to the foreign producers
under a quota.

We have so far considered the terms of trade only in
the light of the elasticities of home and foreign supply and
demand. Any increase in the balance of trade, by which-
ever expedient it is brought about, will lead to an increase
in home incomes and activity. It will therefore raise both
the demand curve for imports and the supply curve of
exports, 1 But the effect of increased incomes in raising
the demand for consumable imports, and the effect of
increased activity in raising the demand for raw materials,
will normally be far greater than the effect of increased
home consumption in reducing the supply of goods avail-
able for export. Increased activity is therefore likely to have
a larger effect in raising the price of imports than in raising
the price of exports, and therefore tells in the direction of
worsening the terms of trade. The presumption that the
terms of trade will deteriorate as a result of a fall in the
exchange rate or of wages is therefore increased, the
deterioration due to a subsidy or to quotas is enhanced,
and the improvement due to a tariff mitigated, by the effect
of increased activity.

The effect of changes in the terms of trade upon income

1 See p. 139.

BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOYMENT 1 67

per unit of employment must be combined with the effects,
discussed above/ of the distribution of home activity
between different groups of industries. The beneficial
effects oT a tariff upon the terms of trade may offset the
deleterious effects of concentrating output in a narrower
group of industries, and in favourable circumstances may
even lead to an increase in income per unit of employment.
Exchange depreciation and wage cuts occupy the inter-
mediate position on both counts; while subsidies and quotas
are the rftost deleterious, on both counts, of all the exped-
ients for increasing the balance of trade.

The change in real wages which is brought about by
the various expedients is not necessarily commensurate
with the change in real income per unit of employment,
for wage earners may consume goods of various types in
different proportions from the average for the country
as a whole, while, in the case of a tariff, the benefit to wage
earners of the expenditure of tax receipts is not necessarily,
or usually, commensurate with the contribution which
they m#ke to them. For a given increase in the balance of
trade, the rise in the home price of export goods is greatest
in the case of a subsidy, and the rise in the price of import
goods, and of home goods which are rival to them, greatest
in the case of tariffs, while a fall in the exchange rate or in
money wages has an intermediate effect upon both sets of
prices (prices being calculated in wages units, in the case
of a fall in money wages). Thus for a country whose export
goods are an unimportant element in the consumption of
wage earners the fall in real wages will be least for a sub-
sidy, greater for depreciation, and greatest for tariffs, while
for a country which exports food-stuffs and imports the
luxuries of the rich the order of preference is reversed.
Quotas, which are commonly applied to agricultural
commodities and so raise the price of food-stuffs, and which
make no contribution to fiscal revenue, bring about the

1 68 ESSAYS IN THE THEORY OF EMPLOYMENT

largest fall in real wages of all the expedients for increasing
the balance of trade.

The various expedients have important effects upon the
distribution of income and activity between industries
within the home country. An increase in the balance of
trade is accompanied by a rise in the home price of export
goods, or of goods which are rival to imports, or of both
together, so that an increase in the balance of trade
increases not only activity, but also income per unit of
output, in the industries concerned in producing these
goods. Now, when the game of beggar-my-neighbour is
being hotly played, these industries suffer a decline in
incomes relatively to the industries which are not subject
to foreign competition, 1 and an improvement in their
situation may be regarded as desirable for its own sake,
apart from any increase in the total of activity and incomes
of the country. This consideration is of particular impor-
tance in so far as it affects agricultural commodities, since
the agricultural community is in general poorer than the
industrial. Any policy which is designed to increase the
exports, or reduce the imports, of agricultural com-
modities has the effect of turning the terms of trade
between agriculture and industry inside the home country
in favour of agriculture, and so of reducing the inequality
in their earnings. Such policies are widely held to be
beneficial, in spite of the fall in the average of real wages
which they necessarily bring about. 2

1 Even in a country so greatly dependent upon foreign trade as Great Britain
these industries occupy much less than half the working population, while the Multi-
plier appears to be normally something in the neighbourhood of 2. Thus a given
decline in employment in the foreign trade industries causes an almost equal absolute,
and therefore a smaller proportionate, decline in employment in the home trade
industries. This is known as 'the problem of the unsheltered industries.*

2 A' fall in the exchange rate, or an all-round reduction in wages, will benefit the
export industries even when they bring about no increase, or even a decrease, in the
balance of trade, while quotas will always benefit the home industries protected by
them, and subsidies the industries which receive them. These expedients may there-
fore be resorted to in certain circumstances entirely for the sake of the industries
concerned, without regard to their effect upon the general level of activity, while
tariffs are often designed for the benefit of particular groups without much regard to
their incidental effect in improving the balance of trade. f

BEGGAR-MY-NEIGHBOUR REMEDIES FOR UNEMPLOYMENT 169

Certain special considerations apply to dach of the four
expedients. We have treated a reduction in wages as being
in general equivalent to a fall in the exchange rate, but
there is one difference between the two which is of the
utmost importance. Even if obligations to foreigners fixed
in terms of home currency are unimportant, internal
indebtedness still has to be considered. A cut in wages
leads to a redistribution of real income in favour of the
fixed-income classes, and an increase^in the burden of
indebtedness within the home country. For this reason
a cut in wages is undesirable so long as any other ex-
pedient will serve, even if it can be brought about smoothly
without the distress and wastage of industrial disputes,
and even if it can be made equal in all industries so as to
avoid arbitrary redistribution of income and activity
between them.

Depreciation of the exchange rate has the disadvantage
of being regarded as a breach of international good faith,
while the apprehension of a fall may have serious effects
upon .the international financial position of the home
country.

Tariffs and subsidies bring well-known political evils
in their train, from which the more general, automatic and
inhuman mechanism of exchange depreciation is compara-
tively free, while tariffs foster monopoly by violently reduc-
ing the elasticity of demand for home goods formerly
subject to foreign competition, and so making the gains of
monopolisation more tempting to the home producers.
Tariffs, it is true, have the advantage that they are selective,
and may be devised in such a way as to bring about the
minimum decrease in real wages for a given increase in
employment, but actually they are not always devised with
this end in view.

All expedients are subject to the objection that they are
calculated to promote retaliation; indeed this is the very
nature of the Deggar-my-neighbour game. Which ex-

170 ESSAYS IN THE THEORY OF EMPLOYMENT

pedient is the feast dangerous from this point of view will
depend upon general political considerations.

When a nation, hard pressed in the game, is determined
to take a trick, the decision as to which suit it is Wisest to
play must be taken in the light of all the considerations
set out above, as they apply to the particular situation
of the nation concerned at the particular moment when the
decision is taken.

From an un-najtionalist point of view all are equally
objectionable, since each is designed to benefit one nation
at the expense of the rest. But there are circumstances
in which a limited indulgence in them cannot be regarded
as a crime. First of all, they may be justified by the plea of
self-defence, and secondly they may be used merely to
cancel out a benefit to the rest of the world that would
otherwise result from the policy of one nation. An increase
in home investment in one country tends to increase activity
in the rest of the world, and measures designed to protect
the balance of trade when home investment increases
merely cause a larger share of the reward of virtue ^to fall
to the virtuous nation, while measures which prote'ct the
balance of trade when money wages rise at home merely
prevent the rest of the world from gaining an advantage,
and leave it no worse than before.

PART IV
INDETERMINACY

THEORETICAL economists usually display an extreme
anxiety to discover 'determinate' solutions of the problems
which they study. Determinacy, for th economist, means
that the* problem is susceptible of analysis by the methods
which he has been brought up to use. When 'causes of
"indeterminateness" are said to be detected this is merely
another way of saying that a new set of determining forces
has been found: forces whose behaviour and manner had
not hitherto been reducible to uniformities'. 1 The problem
is 'indeterminate 5 when the specifically economic factors
(that is, factors the economist has been accustomed to
handle), do not serve to provide a unique solution, so that
'non-economic 5 factors have to be brought into the story
before an answer to the problem can be found. 'Non-
economic 5 factors, such as human error or sentimentality,
political disturbances, the state of confidence in the
currency, or the strategical position of Trade Unions,
are those which cannot easily be fitted into the existing
structure of pure economic analysis.

The first task of the analytical economist is therefore to
discover determinate problems on which to work. Some
writers, brought up in the tradition of 'methodological
pessimism 52 consider this task of such importance that they
attempt no other. After counting up the number of equa-
tions and the number of independent variables involved
in a problefn and triumphantly pointing out that the two
numbers are equal, so that the solution of the problem is
determinate, they are quite content to let the matter rest,
without making any attempt to tell us what the solution is,

1 Kaldor, Review of Economic Studies, February 1934, p. 122.
* See my pUmphlet Economics is a Serious Subject.

171

172 ESSAYS IN THE THEORY OF EMPLOYMENT

Others, taking a less humble view of their function, are
apt to resort to special pleading to show that some problem
which they desire to solve can in fact be completely dealt
with by the methods at their disposal, and to display un-
reasonable hostility to anyone who attempts to maintain
the contrary opinion.

This bias in favour of determinateness is certainly very
natural. If the problems in the real world are not deter-
minate in the economist's sense, predictions about events
in the real world cannot be made by means of economic
analysis, and economics can never become a science.
Each particular eyent can only be studied after it has
occurred, by the methods of the historian. Or, alternatively,
the economist must wait for the development of a science
of social psychology to supplement his own methods of
analysis. To discover too much 'indeterminacy 5 in the
real world would therefore deprive the economist of self-
respect, if not of the means of livelihood. But to discover
too much 'determinacy' in his theoretical system ought to
damage his self-respect quite as much. For it is, after all,
the aim of theory to reflect as closely as possible the
conditions of the real world, and we all know from daily
experience that in the real world many problems are
indeterminate in the economist's sense. My neighbour
owns a small paddock at the bottom of my garden, which
I desire to buy. No one but he can supply what I require,
and no one but I wishes to buy from him. The price which
will be settled between us cannot be predicted by any
purely economic line of reasoning. It will depend on his
avarice or sense of fair play, upon my weakness at bluff
or ability to appeal to his better nature. If economic
theory were to tell me that there is only one price which
can rule, I should know very well that economic theory was
wrong. Similarly with wage bargaining. It is obvious
that the determination of actual wage rates is largely
influenced by non-economic factors. But economists are
apt to resent the suggestion that marginal productivity

INDETERMINACY 173

analysis is not sufficient to give a aeiermmaie answer
to the problem.

For instance, when Mr. J. A. Hobson pointed out that
marginal productivity analysis cannot deal with a situation
in which the amount of labour employed must be altered
in an appreciable proportion or not at all, he was very
severely dealt with by Edgeworth, 1 who delighted to mock
at his ignorance of the differential calculus. Mr. Hobson's
innocence was no doubt an irresistible temptation to Edge-
worth's* wit, but it is necessary for the mathematical
economist to concede to the simple-minded critic that in
the real world the minimum feasible change in the amount
of labour employed is often a large proportion of the
whole, and that when this is the case the principle of
marginal productivity cannot be made to provide an exact
account of the rate of wages that will be paid.

Another example is provided by a criticism by Mr.
Hicks of Pareto. 2 Pareto argues that where the proportions
of the factors are fixed by technical considerations the
principle of marginal productivity will not suffice to deter-
mine the wages of the factors. He gives the example of a
silk factory where an addition to the amount of floor-space
on which the silk can be spread out adds nothing to the
output of silk. 'This, 5 remarks Mr. Hicks, 'is merely silly.'
Now Pareto's example may not be very well chosen.
Probably more space for walking about between the pieces
of silk would be very useful. But clearly the marginal
productivity of any one factor falls off more rapidly as
that factor is increased, the others remaining constant,
the more rigid are the technical conditions. In the limit-
ing case, where only one combination of factors is tech-
nically possible, the marginal productivity of any one
factor falls off infinitely rapidly as that factor is increased,
and the value of the marginal productivity is not uniquely

1 Papers Relating to Political Economy 9 vol. i, p. 19, note.

* * Marginal Produrtivitv and the Principle of Variation,* Economica, February 1932,
p. 86, note.

174 ESSAYS IN THE THEORY OF EMPLOYMENT

determined. In this particular case all ended well, for
Mr. Hicks was able to show that when the 'factors' are
widely defined and when the conditions of supply of factoi^
are brought into the story a determinate value fdr their
prices can be found. But the reader cannot help but suspect
that Mr. Hicks' comment upon Pareto arose from over-
anxiety to discover a determinate solution for the problem.

These examples are concerned with real wages. To turn
to the question of c money wages: economists of the most
divergent opinions agree that in Great Britain between
1926 and 1931 prices were maintained at a relatively high
level by the tenacity of Trade Unions in refusing to accept
wage-cuts in the face of severe unemployment. Yet, in
the theory by which we were brought up to account for
the general level of prices, the bargaining power of Trade
Unions never seemed to appear specifically in the equations.
Of course, a story could be told of how inexorable economic
forces play upon the minds of the Trade Union secretaries
in such a way as to force them to decide upon a certain
line of conduct. But the story was very long, very com-
plicated and very unconvincing.

It is a great merit in the General Theory of Employ-
ment that it allows us to believe that the general level of
prices is determined very largely by arbitrary human
decisions, and yet saves our self-respect by leaving us such
problems as the determination of the level of real wages and
the amount of employment to be discussed by the methods
of pure economic analysis.

AN ECONOMIST'S SEKMUIN

Economics is the dope of the religious people

CONSIDER the case of a man to-day who has an honest
intelligence, a strong social conscience and an independent
income.

His intelligence tells him that he has no particular right
to enjoy a privileged position. 'Right' is a vague phrase.
A doctor has in a sense a right to a mgtor-car because it
makes him do his work better than he could without it.
And if he uses it to visit his friends as well as his patients,
no harm is done to anyone. But our man is too honest to
try to persuade himself that his own comfort really makes
very much difference to the amount of benefit that he does
to other people. His conscience tells him that he would be
doing a good act if he endowed a hospital with his wealth
and worked for his living. But his independent income is
not easy to give up.

He cannot keep all three integrity of mind, a quiet
conscience, and the privileges of wealth. One must be
sacrificed. If he is a saint he sacrifices the wealth but we
will suppose that he is not. If he is a man of no definite
religious creed he can keep his mental honesty and his
income by sacrificing his conscience. He can say C I am a
selfish individual. I don't pretend to have any better right
than anyone else to a comfortable life, but I propose to
enjoy it if I can. 3

But if he belongs to a definite religion this line of escape
is impossible for him. Conscience is more precious than
anything else. Without its approval he can have no peace.
He will have to sacrifice his honesty of mind instead, and
make up arguments to show that it is right for him to be
better off than the majority of his neighbours.

Sfow, it is her that the economist is a godsend to him.

175

i 7 6 ESSAYS IN THE THEORY OF EMPLOYMENT

The economist is a self-appointed expert. It is his business
to know about these things. A man may have an honest
and independent mind and yet take on trust the opinion of
experts on a subject that he has not time to ma f ster for
himself. If the economist tells him it is all right, then he
can keep his integrity, his income and his conscience all
intact.

One of the main effects (I will not say purposes) of
orthodox traditioi^l economics was to fill this want. It
was a plan for explaining to the privileged class that their
position was morally right and was necessary for the welfare
of society. Even the poor were better off under the existing
system than they would be under any other. There is a
significant passage in the reminiscences of Alfred Marshall. 1
As a young man, a mathematician and philosopher, before
he had embarked upon economics, he began to be troubled
by social conscience.

From Metaphysics I went to Ethics, and thought
that the justification of the existing condition of society
was not easy. A friend, who had read a great deal of
what are called the Moral Sciences, constantly said:
c Ah! if you understood Political Economy you would
not say that. 5

Marshall himself did much to break down the doctrine
that no matter how much poverty and distress there may
be it is still true that all is for the best in the best of all
possible worlds. But even in the system of economics as it
was handed down by Marshall the main theme is still the
justification of the existing system.

I will put forward three examples of the products of
traditional teaching: First, the doctrine that increased
wealth of the propertied class brings about an automatic
increase of income to the poor, so that, if the rich were
made poorer, the poor would necessarily become poorer
also. A familiar instance is to be found in Mrs. Marcett's

1 Memorials of Alfred Marshall, p. in*

AN ECONOMIST'S SERMON 177

Conversations on Political Economy, a popular eaVly nineteenth-
century work designed to explain the principles of Political
Econbmy to the masses. Mrs. B. explains to Caroline the
subject of Poor Law Relief:

The greatest evil that results from this provision for
the poor is that it lowers the price of labour; the sum
which the capitalist is obliged to pay as poor rates
necessarily reduces the wages of his labourers, for if
the tax did not exist, his capital boing so much more
considerable, the demand for labour and consequently
its remuneration would be greater . . . x Where there
is capital the poor will always find employment, the
demand for labour is therefore proportioned to the
extent of capital. 2

This argument is based upon the postulate of nineteenth-
century economic teaching, that an increase in the capital
equipment of industry is the result of an increase in thrifti-
ness. An unsophisticated inquirer might wonder how it
comes about that a decline in the demand for consumption
goods, * due to transferring income from the poor who
must spend it to the rich who may save it, can possibly
lead to an increase in the demand either for the labour
or for the capital equipment which is required to make
the goods. But he is silenced by appeal to the mysterious
Law of Economics which teaches that saving is only another
form of spending. This Law, as it happens, entails
that unemployment cannot occur, so that, if it held true
in fact, c the poor 5 would always be fully employed in any
case. Mrs. Marcett's argument that the thriftiness of the
rich is necessary to provide ^employment for the poor is
therefore a violent contradiction. The whol basis of the
Argument lies in a simple confusion of thought.

But that is not the concern of our honest inquirer with
capital of his own. He can take the word of the experts
and rest satisfied. If the experts' arguments do not hold

1 Conversations 01$, Political Economy, p. 164. a Op. cit., p. 130.

178 ESSAYS IN THE THEORY OF EMPLOYMENT

water it is not his business to put them right. And see how
comforting their word is. If poor relief did the poor good,
the awkward question might arise whether he ought not
to contribute to the aid of the poor more than the State
demands, or ought not to vote for a government that would
demand more. But if even poor relief does the poor
positive harm then he can go on enjoying his comfortable
life without distress of mind.

My next example is more modern. By a statistical
calculation, which has had considerable vogue, it is shown
that, if the present national income were equally distributed,
even the poorest \\rould be hardly any better off than they
are now. This leads to the conclusion that it is no good
causing the sacrifice to the rich of redistribution, as it will
only reduce everyone to a dead level of poverty. For my
own part I have never been able to see the moral force of
this argument. If the average is so low, it seems to me
all the more disgraceful to live above the average standard.
But actually this conclusion is found to be of great comfort
to the social conscience of the wealthy.

Now, even on the assumption that the aggregate con-
sumption of the nation remains unchanged, the calculation
is very hard to believe. A calculation in money values has
little meaning a calculation in terms of productive
resources is obviously required and it is hard to believe
that if, for instance, the equivalent of all the domestic
servants who would be dismissed from wealthy households
after the redistribution of income were turned into attend-
ants at nursery schools, the average of welfare would not be
increased. But however that may be, it is impossible to
maintain that the aggregate of consumption would in
fact be no greater than before, for the thriftiness of the
community would be lowered by redistribution, and taking
one year with another, the total waste of resources through
unemployment would be very much reduced. The
aggregate of consumption and the average standard of life
would therefore be raised. The statistical argument, like

AN ECONOMIST'S SERMON 179

so many others which provide dope for the conscience
of the r pious rich, loses its force when the Law that saving
is a form of spending has been removed from the statute
book.

My last example is taken from a letter td The Times of
the Bishop of Gloucester. At the time of the 'National
Crisis' in 1931 it was considered necessary to introduce
emergency measures to balance the budget. In the name
of 'equality of sacrifice 5 an addition was ijiade of 6d. to the
income tax and a cut of 10 per cent, in unemployment
benefit. In 1934, the budgetary position being favourable,
the Archbishop of York published a pleain The Times that
in the event of a surplus, the restoration of cuts in the
allowances of the unemployed should come before any other
concessions, including remission of income tax.

To his letter the Bishop of Gloucester replied as follows: 1

*i. "Unemployment is more than a misfortune for those
who are overtaken by it; it is a curse." So he [the Arch-
bishop of York] tells us. I agree, and therefore it seems to
me to be the first duty of the Government to do all in their
power to reduce the number of unemployed. It is recog-
nized that one of the most fruitful causes of unemployment
is excessive taxation, and in particular a high income tax.
I am convinced that nothing would ease the situation more
at the present time than a substantial reduction of income
tax, for it is very largely a tax on industry. We are none
of us concerned with the discomforts of those, whether
ourselves or others, who are reputed wealthy. But the
experience of history tells us that it is always the poorer
classes who suffer most from excessive taxation. If a man
is compelled to dismiss a certain number of his workmen,
it may diminish his profits or comfort, but it ruins the men
dismissed.

'2. The result is still more the case if funds thus diverted
from industry are used to increase the amount spent in
supporting the unemnloved. for all monev snent in that wav